Kai Guy (NA)
: 1. Fill protection. 2. Premade teams.
> [{quoted}](name=Kai Guy,realm=NA,application-id=3ErqAdtq,discussion-id=NImxoNEe,comment-id=0000,timestamp=2019-11-16T16:12:38.820+0000) > > 1. Fill protection. > 2. Premade teams. 1. Fill protection can be included. A trivial (and sub-optimal) solution is the following: you don't run re-shuffling if the 10-player buckets contains a player with fill protection. The optimal solution is, yet again, to solve a combinatorial problem, although slightly harder than the one without fill protection (this is a combinatorial problem with excluded cases, so the solution space is actually **_smaller_**, but additional checks may be needed). 2. I have already explained this. There are solutions (albeit more sophisticated) for DuoQ as well, which amount to basically a generalization of the described procedure. What I have described in my previous post is the **_simplest case_** (10 SoloQ players, no fill protection), just to give a basic idea. Note that the optimization goal is SoloQ/DuoQ. FlexQ is a totally different queue and so can run the usual system (and suffer from the usual downfalls). But, frankly speaking, Autofill in FlexQ is not as bad as Autofill in SoloQ/DuoQ. Though, but this is a personal preference, I would like FlexQ removed (since it doesn't make much sense anyway) and Ranked-5 reintroduced as a separate queue. And Ranked-5 doesn't need Autofill, so it's fine. In any case, the goal of my post was to show that even the simplest case (i.e. 10 soloQ players without fill protection, which, considering the very high numbers of games being played globally, is still a huge number) shows that Autofill can be optimized with low-complexity algorithms. As a final note, I am not going to argue that Riot should implement the optimal solution to this problem (which still exists and solves all cases, including fill protection and FlexQ). They aren't able to make a Client work, so any non-trivial algorithm is probably going to end up in disaster. But, as a bare minimum, the simplest case should be tackled and solved.
Rioter Comments
IainG10 (EUW)
: > [{quoted}](name=Xintium,realm=EUW,application-id=3ErqAdtq,discussion-id=X12pxYhg,comment-id=0010000000000000,timestamp=2019-11-05T09:48:12.660+0000) > > A note on role-distribution: this is Riot fault, no matter how you look at it. Considering the huge number of high elo players (across the globe), the fact that some roles are consistently viewed as "bad", except by a relatively small percentage of dedicated players, simply means that Riot balancing of that role is flawed. Top Lane seems to be the main issue, nowadays. I do not know the specific role-distribution now, but back in the day, being a support (also called a "ward-bot") just felt miserable, yet a necessity, and Riot publicly admitted that support was the least played role, causing insanely long (30 minutes is a joke, do you suffer from long-term memory loss? 1+ hour Q-time says hello!) Q-times at the very top of the ranked ladder. I strongly agree with your post, but to expand on this bit right here; if you actually stop to think about it, there were ways for Riot to combat this that they either missed due to not really playing the game for pleasure or just missed through sheer ineptitude. Why are they making people learn roles via Autofill in Ranked games, rather than giving incentives to learn in Normals? Why not have a weekly mission to win a Normal matchmade game in the previous week's least popular Ranked role for some not completely useless reward, like a Key fragment or 25 OE, or even an Honor Booster that doubles your received Honor for x games? I honestly thought Role-based Ranks were a good idea on paper, but the implementation not only killed that iteration, it killed any chance of any other iteration because there was not sufficient reason to actually try in your off-role games (be it a lack of carrot or a lack of stick, though IMO it was both). Riot saw that some roles were a punishment and instead of deciding to at least give some compensation (but not LP gain boost, because that would 100% fuck up the system and even I, a Support Main, am against that) and maybe make the role more attractive and therefore more populous, they just decided that everyone should suffer more equally.... This is a serious step on the road to game design by numbers rather than feel, and it kills games almost as hard as being bought out by EA does.
Well said. At this point, it is sheer "ineptitude". But not only in the sense of technical ineptitude, which they publicly admitted ("spaghetti code") anyway, but also in this sense: LC$ above all. P.S. I'm also a Support (I mean, ward-bot) main.
Wolfeur (EUW)
: Autofill *is* needed because otherwise queues for the most popular roles would pile up indefinitely.
> [{quoted}](name=Wolfeur,realm=EUW,application-id=3ErqAdtq,discussion-id=X12pxYhg,comment-id=001500000000,timestamp=2019-11-05T11:35:51.004+0000) > > Autofill *is* needed because otherwise queues for the most popular roles would pile up indefinitely. The basic premise of the whole thread is that we are reasoning on a system for the upper-part of the ranked ladder, not for the milions of mid-elo and low-elo players. We are reasoning on "how to properly do match-making for a relatively small sample". Not on "how to do match-making for everyone". In the "small" sample scenario, the indefinite pile-up simply doesn't happen: only a transitory pile-up happens. The real question is "how long should that transitory phase be ?" Autofill is simply an automated choice on the waiting time. Autofill essentially boils down to: "after N minutes, ignore role preference", where N is a fixed number, say 5 minutes. Since Autofill is totally random, it may even happen (as it actually happened several times, and this is well-known) that you queued up for MidLane and got autofilled Jungle, while a player who queued up Jungle got autofilled Mid. It may happen that they are in the same team. But if they end up in opposite teams, they won't even know that. A simple team re-shuffling would have solved that, avoiding any autofill altogether. Instead of that nonsense, in Full-Role-Preference, you get 3rd or 4th role more consistently and 5th role consistently less often. That's it.
: Here's the big problem with too much player choice in how you queue. Splitting up the queues and making them essentially incassesible to eachother makes for longer queue times and worse game quality for everyone. If people are queueing for a fast game and there are 8 other role-compatible players queuing at the same time you could just put 2 slightly lower role-compatible players or 2 "autofills" in that game and start it up. But now if 3 of those 8 people are in the "1 hour good matches only" queue you suddenly need 5 extra players instead of 2. This decreases game quality for "fast queues". And since fast queuers are constantly being snatched up there's hardly any players to match with those 3 1 hour queuers. It also really fucks with the matchmaking as outside of happy coincidences where fast queuers just happen to be availible and role-compatible at the perfect time they won't be matched with/against slow queuers, essentially creating 2 different ladders. Maybe you suddenly realise that if most people are fast queueing that by slow queueing you'll dodge most of the best and most practiced players, being able to earn Elo against worse players and inflate your rank. At high Elo the absolute last thing you ever want to do is segregate the playerbase, you want all the best players playing against eachother whenever possible. As an added negative it would also be insanely abusable for duo boosting/wintrading, it would be so much easier to snipe games together and either get a "free" duo or end up against eachother and decide on who's gonna trade the win.
> [{quoted}](name=SMURFF NAME HERE,realm=EUW,application-id=3ErqAdtq,discussion-id=X12pxYhg,comment-id=00100000000000000000,timestamp=2019-11-05T11:47:55.670+0000) > > Here's the big problem with too much player choice in how you queue [...] I strongly believe everything you have written: 1. already happens now; 2. can be solved by fine-tuning. The queues are not "inaccesible" to each other, I don't know how you reached that conclusion. Queue times are also not necessarily longer. But game quality is significantly improved (not worsened). Probabilistically speaking, "slow-queuers" (the ones who select a high Max_Timer) are realistically only being matched with "fast-queuers" (low Max_Timer) with the same MMR range. To match 10 slow-queuers in a dynamic environment, the probabilities are astronomically low, especially considering that those 10 are likely all in Q for Mid Lane, so the system won't match them together until they expire the Max_Timer and then start looking at 2nd, 3rd, 4th and 5th roles. The end result is that the "slow" and "fast" ladders will constantly get mixed. Note that players who select 1 hour as Max_Timer are willingly doing that: the system isn't forcing anything upon them. You say that they are matched against better practiced players. That's impossible, since better players have higher MMR and the system, to begin with, just discarded those. The whole purpose of having 1 hour Q-time is to get inside the appropriate MMR bracket, not only to get your primary role, which likely won't happen anyway if the primary role is already "saturated" by the playerbase (with or without this system, this is a fact which is system-independent). If the MMR bracket is essentially random (due to MMR being a random number), like it is now, that's a different story and I may be inclined to agree with you, but it is not the fault of a match-making algorithm. Finally, I think we already established that waiting 5-10 more minutes is worth if the game quality (i.e. the MMR bracket) is better. "Rank inflating" already happens now. Hello, negative winrates high elo players! How have you been? Rank inflating, which is simply a by-product of imperfect MMR matchmaking, is unavoidable in any system, except the "infinite-Q-time" system (are you happy to wait 1 year for a game? yeah, have fun waiting for that "perfect" match), which is trivially bad. Do not think that Autofill magically makes you immune to rank inflating: that's, mathematically speaking, nonsense.
Wolfeur (EUW)
: It's funny how people still don't realize that Autofill is a literal *necessity* for matchmaking to work…
> [{quoted}](name=Wolfeur,realm=EUW,application-id=3ErqAdtq,discussion-id=X12pxYhg,comment-id=0015,timestamp=2019-11-05T10:12:12.083+0000) > > It's funny how people still don't realize that Autofill is a literal *necessity* for matchmaking to work… Not exactly. To be more precise, it is necessary to provide a solution to the "Queue Time vs Game Quality" trade-off problem. An algorithm dealing with match-making needs to solve this (otherwise no matches can be made). Autofill is _one_ solution. It is not the _only_ solution, neither the _best _ solution. Full role preference is another solution, as I explained in my previous post.
: Adding a third role would reduce the number of autofills considerably but it probably wouldn't help much with game quality. I see people say they got "autofilled" when they get their secondary role and just look at the number of "suggestions" to queue for only one role with """"slightly"""" increased queue times. I'm pretty sure that if a third role was added people would endlessly bitch and moan about it and how they're constantly not getting their primary etc. etc. Though there's pretty much no reason not to do it since it's strictly better than random autofill. At that point you could also just give people a full role preference list making sure that outside of really freak occurrences with extremely small player pools(like challenger) you never get your 4th/5th role. And role distribution is pretty bad, mid is heavily over-represented(which means midlaners get autofilled to other roles all the time) while top is definitely under-represented. I have, no joke, not been autofilled or even got my secondary(mid) once in like 100 games when I queue top and lane against autofilled players all the time. It's obvious that the system is starved for people queuing top.
Disclaimer: I am not high elo myself, but I know (in-real-life) some Diamond+ players. > [{quoted}](name=SMURFF NAME HERE,realm=EUW,application-id=3ErqAdtq,discussion-id=X12pxYhg,comment-id=001000000000,timestamp=2019-11-05T06:44:06.214+0000) > >Though there's pretty much no reason not to do it since it's strictly better than random autofill. At that point you could also just give people a full role preference list making sure that outside of really freak occurrences with extremely small player pools(like challenger) you never get your 4th/5th role. This. High Elo needs full role preference. Full role preference is strictly better than Autofill. The fact that it isn't a magic bullet doesn't mean it shouldn't be implemented, leaving a strictly worse system (aka autofill) in place. A note on role-distribution: this is Riot fault, no matter how you look at it. Considering the huge number of high elo players (across the globe), the fact that some roles are consistently viewed as "bad", except by a relatively small percentage of dedicated players, simply means that Riot balancing of that role is flawed. Top Lane seems to be the main issue, nowadays. I do not know the specific role-distribution now, but back in the day, being a support (also called a "ward-bot") just felt miserable, yet a necessity, and Riot publicly admitted that support was the least played role, causing insanely long (30 minutes is a joke, do you suffer from long-term memory loss? 1+ hour Q-time says hello!) Q-times at the very top of the ranked ladder. Full role preference can mitigate (careful: not "solve") all of this. And how can you prevent 1+ hour Q-time at the very top? Simple: add a MAX_TIMER (say, 30 minutes), which players can see while they are queueing. After (Q-time > Max_timer), two things happen: 1. The player-in-queue gets an alert, saying that Q-time exceeded the max_timer; 2. This means that the MMR range for the match-making will get slowly extended as time passes. If the player is fine having to play against a larger MMR range, he just needs to keep waiting. Most likely, after 5-10 more minutes (35-40 minutes total Q-time), he will find a match. Not "the" perfect match, but likely a somewhat fair match nonetheless (MMR isn't perfect anyway, it's already an estimate with a bad variance, so...). If the player isn't fine and really, like _really_ wants that "perfect" MMR match, he will wait 1+ hour. His choice. When you get to Grand Master (or even Master, depending on the average number of active players, but only Riot has the data to decide this) and above, let players pick the MAX_TIMER between the following options: 1. Zero: you are willing to play a match where the algorithm will search with increasingly large MMR from the very beginning - you don't care playing against Faker in midlane as a GM, since you think getting trashed by him is a learning experience in itself - and one day you may actually snatch a win against him! who knows; 2. 10-20-30-40-50 minutes: the usual deal, with varying degrees of allowed waiting; 3. Infinite: you don't care about waiting 1 day or more, just get that perfect match. After a transitory phase, where players will learn (at their expenses) that trying to abuse the settings only to get their 4th role after 3 hours of Q-time wasn't such great of an idea, the situation will just be a strict improvement compared to now. **Conclusion**: Full role preference is strictly better than Autofill (fact #1). Q-Time and Game quality have an inherent trade-off (fact #2) and this is exacerbated at the top of the ladder (fact #3). Instead of having the system arbitrarily decide what is the priority and having autofill ruin games, let the players pick that. At most, the players will ruin themselves. But after a transitory phase, where players trying to abuse the system will get trashed by it, the situation will significantly improve. _**The numbers/ranks given here are purely made up**_. I don't know if MAX_TIMER needs to be 20 minutes or 30 minutes. I don't know if it needs to be tunable from Diamond 1 or from Grand Master. Only Riot, having full datasets, can make these precise fine-tunings. But if you have the full picture, iteratively tuning this system is very doable. And it can only get better.
Saezio (EUNE)
: Cool example. Yeah that game is exactly what I am talking about. Who was funnelled in that game? That's not the typical funnel where you have stuff like yi+taric/nunu+karthus tag teaming jungle and mid. Neither the xayah/kaisa in mid with smite and double supports in the team. That's a perfect example to shut me up. (DAMN IT!) Ok maybe I was wrong and in some cases Jungler was indeed the second person in their team to hit level 6. I still think lowering the jungle xp EVEN MORE than it is now is bad. And I still think what Riot should have done is, lower the FIRST CLEAR XP SIGNIFICANTLY, slightly increase gold on it (so that counterjungling in the first minutes is viable, so jglers are KEPT in the jungle OR risk to lose much) and at the same time INCREASE the difficulty of doing the camps significantly, (maybe spawn the camps a bit later so the jgler has to clear alone without help from lanes?) I think lanes should be left alone for a minimum of 4-5minutes in the standard way to play the game. So that the laners have more agency in their own lane. Then increase jungle XP in 2nd 3rd 4th clear etc to make up for the lost XP. I think Jungler's role should be normalised during the entirety of a 45 minute game. And not be SUPER impactful early, but have little to no agency late outside of being the one to smite the objectives.
> [{quoted}](name=Saezio,realm=EUNE,application-id=3ErqAdtq,discussion-id=LQAEEnFx,comment-id=000200010001000000010000000000000000000000000000,timestamp=2019-10-18T11:32:33.007+0000) > > I still think lowering the jungle xp EVEN MORE than it is now is bad. Same, as already explained. > I think lanes should be left alone for a minimum of 4-5minutes in the standard way to play the game. So that the laners have more agency in their own lane. Agreed.
Saezio (EUNE)
: So, you are saying there are games in that meta, where the jungler powerfarmed and got to level 6 as second person in their team? I really don't remember any games like that. I remember ADCs picking multiple lanes of farm because top laners were tanks that didn't require a high budged to function. If you remember any games where JG was the second person to hit level 6 in their team without something hilarious happening (like the JG being 8-0 by that point) I would be happy to accept I was wrong. But the point stands, JG was always behind at least 2 people in their team in XP at level 6. And if we check level 13-14 for example, JG is almost always second to last in XP at that point being only above support.
> [{quoted}](name=Saezio,realm=EUNE,application-id=3ErqAdtq,discussion-id=LQAEEnFx,comment-id=0002000100010000000100000000000000000000,timestamp=2019-10-17T17:39:39.992+0000) > > So, you are saying there are games in that meta, where the jungler powerfarmed and got to level 6 as second person in their team? Did I say that? > I really don't remember any games like that. I remember ADCs picking multiple lanes of farm because top laners were tanks that didn't require a high budged to function. You are confusing Lane Swap meta with Gold Funneling meta. A quick refresher is needed. In Spring 2015, these were the most prevalent toplaners (data combined from EU and KR, the regions who truly played meta back then): 1. {{champion:150}} (85 games) 2. {{champion:57}} (82 games) 3. {{champion:68}} (57 games) 4. {{champion:127}} (35 games) 5. {{champion:117}} (28 games) This was also the Golden Age of Kalista, being the literal best ADC in the game, sitting alone on the throne of S+ tier (meta-warping), a whole tier above everyone else. Gnar and Maokai shared S+ tier in top. Other roles had 2-3 "viable" (S tier) champions per role, but no clear winner. Lane Swap worked like this. You send your botlane top and your toplane bot. If the other team just plays normally, you have the following situation: 1. TopLane: your ADC + Supp vs enemy Top; 2. Jungle as normal 3. Mid as normal 4. BotLane: your Top vs enemy ADC+supp Now, the lanes with 2 people start to perma-freeze the wave (which was actually possible back then). So you get both toplaners having literally 0 cs and they were zoned out of EXP range for most of the time. In this situation, if mid and jungle just power-farm and nothing happens on the map, you get Mid as first level 6 and Jungle as second level 6. This is "basic Lane Swap". However, this is also an ideal scenario. This is not going to happen so perfectly clean, for a simple reason: toplaner is too vulnerable to 4-man ganks, which is the next macro-game step to make. Tank or not, a properly executed dive will kill him. Since teams were aware this could happen, they shaked things up by adding variants: support going with top or just perma-roaming to track enemy jungler and get vision (and this is the reason why ADC hits level 6 first, not anything else), other teams answered lane swap with another lane swap, creating "australian lanes" (both botlanes go top, both toplaners go bot), top laners farming alongside jungler in the Jungle, junglers lane-farming, etc. > But the point stands, JG was always behind at least 2 people in their team in XP at level 6. This was not your point. Changing a statement (no matter how slightly) changes if it's true or not. This was your point: > I don't remember a single meta where junglers were even hitting level 6 before mid/top. To which I answered: Lane Swap meta. And note that top was actually a solo-lane (there were games with 3 solo-lanes: top, mid and bot, with support perma-roaming without getting any EXP until later in the game). It just got _starved_, as I explained before. Starved does not mean only gold, but also EXP. You are saying my example is invalid because ADC gets level 6 before Jungle. To which I answered that it is irrelevant, since your point was about mid/top, not about "who gets level 6 first". I am not wasting time trying to find a game where the "ideal" Lane Swap described above was actually played. > If you remember any games where JG was the second person to hit level 6 in their team without something hilarious happening (like the JG being 8-0 by that point) I would be happy to accept I was wrong. [FNC vs GAM](https://www.youtube.com/watch?v=5grmasNa15c), but now you are just gonna dismiss it as being "gold funneling" towards the jungler and so invalid, since "gold funneling" is "hilarious" from your point of view (for me, it is not: it was a perfectly legitimate strategy during that time). This is the problem when discussing with ever-changing statements and vague terms ("hilarious").
: I cant believe it’s an unpopular opinion now, but riot should continue balancing around pro play.
There is a deeper problem: pro-play and SoloQ are different _**games**_. They happen to share visuals and some gameplay features. But that's it. In an ideal world, balancing around pro-play would work optimally, since "SoloQ" would simply not exist and be entirely substituted by Ranked Premade5, with vocal chat, and a coach teaching the players macro-game concepts. But this is just a hypothetical, no more. Some champions are too team-relient to work in SoloQ, but they are OP when pros use them. Other champions are very effective in SoloQ, but have little impact on pro-play. This is essentially unfixable. This is a case where Riot stands in a precarious position: on one hand, Riot needs to balance for the majority of the players, not to alienate/frustrate them, ultimately driving them away, which would lead to economical disaster. On ther other hand, Riot needs to balance for pro-play in order to avoid pro-play becoming stagnant, boring... and also an economic failure (no one would buy tickets anymore if pro-play matches were all essentially equal to each other). In a certain sense, they need to balance for both. But this may lead to conflicting goals and you run the risk of making everyone unhappy.
Saezio (EUNE)
: You do realise in the game you linked that happened because both teams decided to give their ADC solo lane EXP instead of their top laner right? (like the majority of that meta) Both mid AND ad carries are level 6 before their junglers. Do I really have to point that out? Can we really not think for ourselves that a game where the top laner is sacrificed for their ADCs experience it's obviously mid+ adc that have to be taken into account and not top? I am really disappointed rn.
> [{quoted}](name=Saezio,realm=EUNE,application-id=3ErqAdtq,discussion-id=LQAEEnFx,comment-id=00020001000100000001000000000000,timestamp=2019-10-17T14:57:35.098+0000) > > You do realise in the game you linked that happened because both teams decided to give their ADC solo lane EXP instead of their top laner right? (like the majority of that meta) > > Both mid AND ad carries are level 6 before their junglers. > > Do I really have to point that out? Can we really not think for ourselves that a game where the top laner is sacrificed for their ADCs experience it's obviously mid+ adc that have to be taken into account and not top? Do you realize that the game I linked is just one of _**many**_ examples of what that meta was? Perhaps you are missing the bigger picture. The whole point of that strategy was to starve enemy toplaner. Your statement was that toplane always hit level 6 before Jungle, in any meta. I just proved to you that this simply wasn't the case: there are conditions under which Jungle hits level 6 before top. Your statement was factually wrong in the Lane Swap meta. In a "ideal" (but for some aspects simplified) Lane Swap scenario, ADC doesn't necessarily need to have solo lane EXP, that was an additional variant. The main thing was to keep toplane starved. So you essentially have: 1. ADC and Support sharing EXP; 2. Toplane having little to no CS; 3. Jungle hitting 6 before top. This is the core. ADC having solo-lane EXP is just a variant on top of that. You don't need that for the core strategy to function. It is just an added benefit, which was conveniently abused (due to how the meta was). So, if both teams decided to have ADC+Supp share EXP, ADC would not have solo-lane EXP and your whole argument would be invalid.
Anu3isII (EUNE)
: You play as usual? The whole concept of the change is to nerf Junglers' EXP gain by nerfing both possible sources of EXP (farming and ganking). If they nerfed only farming EXP, you would see only Ganking Jgs camping lanes and soaking EXP from them, completely avoiding the change at most, while Farming Jgs would essentially be awful picks, exactly as Saezio said. The change is supposed to affect both play styles.
> [{quoted}](name=Anu3isII,realm=EUNE,application-id=3ErqAdtq,discussion-id=LQAEEnFx,comment-id=0002000100000000000000000000,timestamp=2019-10-17T11:43:55.000+0000) > > You play as usual? The whole concept of the change is to nerf Junglers' EXP gain by nerfing both possible sources of EXP (farming and ganking). > > If they nerfed only farming EXP, you would see only Ganking Jgs camping lanes and soaking EXP from them, completely avoiding the change at most, while Farming Jgs would essentially be awfuk picks, exactly as Saezio said. The change is supposed to affect both play styles. It doesn't work that way. There is an [informative video explaining why](https://www.youtube.com/watch?v=3eLBgmDblHs). Nerfing Jungle EXP leads to ganking become more valuable. And a nerf to EXP soaking doesn't compensate all the advantages that a successful ganks gives you (which are more than simply EXP and gold), so ganking will still be preferred strategy.
Saezio (EUNE)
: Disregarding levels 2 and 3, jungle could never out level solo lanes. I don't remember a single meta where junglers were even hitting level 6 before mid/top. Jungle does have less gold than adc, like every single game. Maybe first clear jg has a slight edge, but except for cases where jungle was stupidly fed or adc stupidly feeding, adc was always more gold than JG at 15 mins. Even at 10 mins most times.
> [{quoted}](name=Saezio,realm=EUNE,application-id=3ErqAdtq,discussion-id=LQAEEnFx,comment-id=000200010001000000010000,timestamp=2019-10-16T19:42:14.767+0000) > > Disregarding levels 2 and 3, jungle could never out level solo lanes. I don't remember a single meta where junglers were even hitting level 6 before mid/top. You are wrong regarding toplane. Junglers were able to hit level 6 before top during the Lane Swap meta (Season 5), where teams actively tried to starve toplane from CS. When it worked, top was 1-2 levels behind jungle. One example: in [this game of FNC vs UOL](https://www.youtube.com/watch?v=LqgZADsNevo), both FNC Reignover (Jungle) and UOL Kikis (Jungle) are level 6, while FNC Huni (top), which was successfully starved, is still level 5 (note that, as I said, this applies only when the starving strategy was successfully executed, which didn't happen in the case of UOL top, since FNC failed to starve him). You may be able, by searching more accurately, to find games where both toplaners were successfully starved and Junglers were able to hit level 6 before them. Also, disregarding levels 2 and 3 is really bad, since that's where Junglers had their peak. Those are the problematic levels and that's where the focus needs to be put. This is also explained in the video I linked in my previous post.
: Good man i like how you took intiative to not only fix but explain your post reasonably. Upvoting for that,
> [{quoted}](name=Chainman3,realm=NA,application-id=3ErqAdtq,discussion-id=WkY06ebA,comment-id=0000000000000000,timestamp=2019-10-17T03:00:47.109+0000) > > Good man i like how you took intiative to not only fix but explain your post reasonably. Upvoting for that, Thanks.
macspam (NA)
: Why is further reducing jungle XP a good thing?
Favoring hyper-ganking strategies is one of the worst possible outcomes for everybody (junglers included). If you didn't already, watch this very useful and informative video about [How Riot Ruined the Jungle](https://www.youtube.com/watch?v=3eLBgmDblHs).
Tobykachu (EUW)
: Nerfing Jungle Experience... Again
This video is quite useful regarding Jungle and it explains many things: [Vertical Jungling](https://www.youtube.com/watch?v=3eLBgmDblHs).
: > [{quoted}](name=Xintium,realm=EUW,application-id=3ErqAdtq,discussion-id=H9QIub3t,comment-id=00010001,timestamp=2019-10-15T13:01:33.278+0000) > > Balancing is an iterative procedure. Your objective is not to "get it right" on the first few attempts. But you do need a reliable and systematic iterative scheme, which, at least theoretically, should converge asymptotically. This is exactly what we have not seen. Riot's balancing framework also has some significant flaws, such as not adjusting winrate with banrate in the case of high ban rates, except with a single correction at 7x ABR. All the stats are actually pretty hard to interpret by themselves. Ban rates are often due to preference and can be very influenced by bias (such as Zed seemingly always having a high ban rate regardless of his state). Win rates are affected by whether a champion is picked mostly as a counter, if it's a versatile champion, if the skill floor is low or high, if the pick rate is high or low and dozens of other factors. There's never going to be a purely objective system for balancing the game because there are just too many factors at play and it keeps changing so much. It's not a finished game like chess.
> [{quoted}](name=astralwit,realm=NA,application-id=3ErqAdtq,discussion-id=H9QIub3t,comment-id=000100010001,timestamp=2019-10-16T03:09:51.597+0000) > > All the stats are actually pretty hard to interpret by themselves. No metric should be looked at in isolation, as if it existed in a vacuum. This is precisely what I claimed in another post. > Ban rates are often due to preference and can be very influenced by bias (such as Zed seemingly always having a high ban rate regardless of his state). As I have already noted, it depends on which banrate you look at. Banrate is not merely a number, it's much more than that. Certainly you won't learn anything useful looking at banrates in Iron/Bronze. But, perhaps, looking at very high elo and pro-play banrate, you can observe all those biases fall-off dramatically. You can also plot banrate w.r.t. (with respect to) Rank and/or w.r.t. MMR, patch-cycles, etc. These are all useful data points. If champion X (please, let us avoid specific champion examples, they only derail the discussion and quickly become off-topic, also they are debatable) is "hated" and has 25% banrate in plat+, but next patch he dramatically rises to 50% banrate plat+, maybe, just maybe, he needs to be looked at more closely. > Win rates are affected by whether a champion is picked mostly as a counter, if it's a versatile champion, if the skill floor is low or high, if the pick rate is high or low and dozens of other factors. Statistics were developed to tackle problems much harder than LoL will ever get. Big Data was not born yesterday, neither was Robust Statistics. But these techniques are not your run-of-the-mill winrate or other "average Joe can read them" stat. They need a lot of effort, time, data analysis and a huge amount of mathematics. The end-result may also be unreadable for non-mathematical adepts. But a proper statistician/data-scientist can in fact read them and knows what, at any given point, the next steps are. > There's never going to be a purely objective system for balancing the game because there are just too many factors at play and it keeps changing so much. It's not a finished game like chess. 1. The changes are not random (although with Riot infamous capability for coding they partially are). They are introduced by Riot devs. 2. You can revert _any_ change you make. That's a **_huge_** advantage. 3. While there are "many" (not really) factors at play, there are also huge datasets available. These factors at play, for the most part (if not totally), are well-known and can be accounted for. 4. There is not going to be a "purely objective" system for balance because "balance" is ill-defined (or, at best, arbitrary), not for any other reason. Ask 10 guys what balance means precisely and you will get 10 different answers, some of them incompatible to each other. But you can set well-define _balance goals_ and try to iteratively reach them. This is objectively doable.
Pika Fox (NA)
: > [{quoted}](name=Xintium,realm=EUW,application-id=3ErqAdtq,discussion-id=H9QIub3t,comment-id=000100010000000000000000,timestamp=2019-10-15T16:06:42.418+0000) > > What is "fotm", as you call it, often needs to be looked at. > > Most importantly, you are talking of bans/banrate relating to _balance_. I invite you to carefully re-read my original post and notice that my main point was that banrate relates to **_winrate_**, as in: > > and in my previous post > > Unless you are using the words "balance" and "winrate" interchangeably, there are some important differences between the two. Fotm is rarely out of line, just fotm. Perceived power and actual power are rarely the same (ie yasuo)
> [{quoted}](name=Pika Fox,realm=NA,application-id=3ErqAdtq,discussion-id=H9QIub3t,comment-id=0001000100000000000000000000,timestamp=2019-10-15T16:14:50.249+0000) > > Fotm is rarely out of line, just fotm. Perceived power and actual power are rarely the same (ie yasuo) I fully agree that perceived power and actual power are practically never the same. I need to ask what you mean precisely by "fotm". Do you mean a champion with high pickrate in a given "month" (patch cycle) ? Because pro-play could not care less about pickrate/popularity. Summarizing, you are saying that: 1. Pro-play banrate is not useful for balance; 2. People ban what they perceive as powerful and/or what they hate; 3. Combining 1 and 2, banrate is rarely useful for balance. I believe that you can, quite trivially, find counter-examples for 1 and 2. The fact that Yasuo isn't one of them doesn't validate the statement. I believe this is where the mistake lies. To be more precise: some champions, like Yasuo, cannot be judged for balance by banrate. But other champions (IMHO, the majority) can.
Pika Fox (NA)
: > [{quoted}](name=Xintium,realm=EUW,application-id=3ErqAdtq,discussion-id=H9QIub3t,comment-id=0001000100000000,timestamp=2019-10-15T13:42:11.842+0000) > > I believe there was a misinterpreting of my post. > > The "banrate" of my previous post is not the banrate across all elos, nor is the banrate in plat+ (or whatever). Those are essentially "unfiltered" metrics affected by all sorts of biases, like systematic hate of a champion. It was used in a very general sense and the specific way of "filtering" banrate depends on the specific balancing problem. For example, taking Yasuo, you may "filter" banrate looking only at the top X% (say, 1%) of the ladder, ignoring the high banrate elsewhere. > > If the top of the ladder still shows a significant banrate, it means Yasuo needs to be looked at more closely (not necessarily "nerf him"). > > However, if you think banrate has no impact on balance, you are indeed wrong. Pro-play banrate, for example, is very valuable. Like all metrics, it needs to be looked at alongside other metrics, never alone. And sometimes it needs to be filtered/normalized/etc. > > What I was actually referring to regarding Riot balance framework is a different matter altogether, which I explained in another post. It is due to the fact that winrate is affected by banrate (which involves confidence intervals, sample size, effective sample size and other things). And Riot decided that a single correction at 7x ABR (an arbitrary number multiplied by 7% plus or minus an essentially random number) was indeed "the correct way" to adjust for banrate affecting winrate. Spoiler: it doesn't work. Pro play banrate is almost entirely targeted bans or counter bans, or simply what is fotm currently. Rarely are bans even remotely tied to balance.
> [{quoted}](name=Pika Fox,realm=NA,application-id=3ErqAdtq,discussion-id=H9QIub3t,comment-id=00010001000000000000,timestamp=2019-10-15T15:18:11.178+0000) > > Pro play banrate is almost entirely targeted bans or counter bans, or simply what is fotm currently. Rarely are bans even remotely tied to balance. What is "fotm", as you call it, often needs to be looked at. Most importantly, you are talking of bans/banrate relating to _balance_. I invite you to carefully re-read my original post and notice that my main point was that banrate relates to **_winrate_**, as in: > such as not adjusting winrate with banrate in the case of and in my previous post > winrate is affected by banrate Unless you are using the words "balance" and "winrate" interchangeably, there are some important differences between the two.
Pika Fox (NA)
: > [{quoted}](name=Xintium,realm=EUW,application-id=3ErqAdtq,discussion-id=H9QIub3t,comment-id=00010001,timestamp=2019-10-15T13:01:33.278+0000) > > Balancing is an iterative procedure. Your objective is not to "get it right" on the first few attempts. But you do need a reliable and systematic iterative scheme, which, at least theoretically, should converge asymptotically. This is exactly what we have not seen. Riot's balancing framework also has some significant flaws, such as not adjusting winrate with banrate in the case of high ban rates, except with a single correction at 7x ABR. People ban yasuo, a completely balanced champion. Bamrate has no impact on balance.
> [{quoted}](name=Pika Fox,realm=NA,application-id=3ErqAdtq,discussion-id=H9QIub3t,comment-id=000100010000,timestamp=2019-10-15T13:06:45.788+0000) > > People ban yasuo, a completely balanced champion. Bamrate has no impact on balance. I believe there was a misinterpreting of my post. The "banrate" of my previous post is not the banrate across all elos, nor is the banrate in plat+ (or whatever). Those are essentially "unfiltered" metrics affected by all sorts of biases, like systematic hate of a champion. It was used in a very general sense and the specific way of "filtering" banrate depends on the specific balancing problem. For example, taking Yasuo, you may "filter" banrate looking only at the top X% (say, 1%) of the ladder, ignoring the high banrate elsewhere. If the top of the ladder still shows a significant banrate, it means Yasuo needs to be looked at more closely (not necessarily "nerf him"). However, if you think banrate has no impact on balance, you are indeed wrong. Pro-play banrate, for example, is very valuable. Like all metrics, it needs to be looked at alongside other metrics, never alone. And sometimes it needs to be filtered/normalized/etc. What I was actually referring to regarding Riot balance framework is a different matter altogether, which I explained in another post. It is due to the fact that winrate is affected by banrate (which involves confidence intervals, sample size, effective sample size and other things). And Riot decided that a single correction at 7x ABR (an arbitrary number multiplied by 7% plus or minus an essentially random number) was indeed "the correct way" to adjust for banrate affecting winrate. Spoiler: it doesn't work.
: Even with "precision and calculation" it will go wrong because a few devs and play testers will never outwit millions of players on the live servers. Something that looks bad to their internal servers and PBE might end up being broken and Something that looks broken might end up being bad on the live servers.
> [{quoted}](name=Abibyama II,realm=EUW,application-id=3ErqAdtq,discussion-id=H9QIub3t,comment-id=0001,timestamp=2019-10-15T09:08:36.304+0000) > > Even with "precision and calculation" it will go wrong because a few devs and play testers will never outwit millions of players on the live servers. > > Something that looks bad to their internal servers and PBE might end up being broken and Something that looks broken might end up being bad on the live servers. Balancing is an iterative procedure. Your objective is not to "get it right" on the first few attempts. But you do need a reliable and systematic iterative scheme, which, at least theoretically, should converge asymptotically. This is exactly what we have not seen. Riot's balancing framework also has some significant flaws, such as not adjusting winrate with banrate in the case of high ban rates, except with a single correction at 7x ABR.
: maybe an ELO system could work for Flex?
> [{quoted}](name=ZackTheWaffleMan,realm=NA,application-id=3ErqAdtq,discussion-id=mFQcnAA3,comment-id=000800000000,timestamp=2019-10-14T18:16:51.853+0000) > > maybe an ELO system could work for Flex? It can work in the case of premade 5. In that case, you may treat a "team" as a single (virtual) "player", assigning LP gains/losses to teams themselves (not to players). In such a case, you could have a team-ladder, completely separated from other ladders, like "Ranked 5". For example, you may have Team A with MMR 2000 (something like Plat) and Team B with MMR 1500. Note: the fact that Team A has MMR 2000 does not imply that every single player has, individually, MMR 2000. This is just the "uniform" approximation, it may or may not be true. Of course, this would not resolve the issues of ranking a player in SoloQ, since using its "team-rank" would basically amount to the same approximations that are made in the case of MMR, and you would be back to square one.
Saezio (EUNE)
: You think pyke is a support that does a lot of damage? Just saying, there are 27 supports that do more. But he does indeed average 1,1k damage more than Janna
> [{quoted}](name=Saezio,realm=EUNE,application-id=3ErqAdtq,discussion-id=WkY06ebA,comment-id=00000000,timestamp=2019-10-14T17:16:04.419+0000) > > You think pyke is a support that does a lot of damage? > > Just saying, there are 27 supports that do more. But he does indeed average 1,1k damage more than Janna No no, it was just an example. I simply inserted Pyke alongside Brand/Zyra, but only to differenciate it from Janna. Nothing more than that. EDIT: Since it may be misleading, I will edit my post and remove Pyke.
Cosnirak (NA)
: I don't think they made any big changes to matchmaking. The matchmaking was always horrendous, and now the game is too, so things are indeed worse and the matchmaking is indeed awful but correlation doesn't equal causation. You can't use an ELO system in a team game where teammates change every game and expect good results. And yet game after game keeps doing it...
> [{quoted}](name=Cosnirak,realm=NA,application-id=3ErqAdtq,discussion-id=mFQcnAA3,comment-id=0008,timestamp=2019-10-14T00:35:36.987+0000) > > I don't think they made any big changes to matchmaking. The matchmaking was always horrendous, and now the game is too, so things are indeed worse and the matchmaking is indeed awful but **correlation doesn't equal causation. You can't use an ELO system in a team game where teammates change every game and expect good results**. And yet game after game keeps doing it... Exactly (emphasis mine). Though, I fear many have skimmed your post, due to not being able to really grasp the mathematical reasons behind it.
: Simple fix(s) to make match making better in ranked.
I agree with the spirit of your post, Free Roll, i.e. judging everyone based on their performance, instead of win/lose. But the problem is that, despite your title, there is no simple metric to do that. And every metric can be "gamed". To assign a numerical value to "performance", you need what is technically known as a "performance score", also called score function or simply _score_. A score is made up of several _metrics_, such as: damage to enemy champions, KDA, CS, etc. This is where things get complicated. Let me give you an example. A Janna support will never have the same "damage to champions" as a Brand/Zyra/... support. So, when you factor in the metric "damage to enemy champions", you are implicitly favoring Brand/Zyra/... supports. To actually evaluate the performance of Janna, you would need the metric "damage prevented to allies" (via shields). Suppose you include such a "shield metric" into your score. Now, you have a dilemma: if the "shield metric", for your Average Janna, gives an overall worse evaluation with respect to "damage to enemy champions" for your Average Brand (support), you are implicitly favoring the Average Brand support. Swap Janna/Brand and you get the opposite. Getting them precisely equal, so that the Average Janna and the Average Brand are on equal footing, is hard. Multiply this process _for every single champion in the game_ and it gets _insanely hard_, if not outright impossible. You may need to fine-tune metrics for every single champion in the game in order to have a score which is actually representative of the performance, not just a random number. This is also the same reason why S+, S, S-, A+, etc. are sometimes completely off. By the way, if your score functions will get figured out by the community (and, sooner or later, it will) you are in trouble, as champions and ways of playing favored by the score will get abused to hell. And that may become a big problem. I still agree with the spirit of what you say. But there is no such a thing as a "simple" fix for ranked match-making.
Xavanic (NA)
: > [{quoted}](name=Keiaga,realm=NA,application-id=3ErqAdtq,discussion-id=9ccYZ9Xs,comment-id=000400000000,timestamp=2019-09-17T05:37:34.638+0000) > > It neans hes statistically and factually not overpowered. Analytics are flawed at best, there is so many extra variables that i would never rely on a number on a site to decide if someone is op or not, everyone bitches and bans yasuo but yet he has a 48% winrate, im not saying trynd is op, what im saying is taking some numbers off a website and saying said champ is op is bullshit, jax is an example of this, same with renekton, their both insanely potent and frustrating to play against right now because of SoS, but neither has a fairly decent winrate due to the META not really favoring diver's, this is why their win rates are meh....once we enter a meta that favors fighters/skirmishers....there is a good chance trynd will become op
> [{quoted}](name=Xavanic,realm=NA,application-id=3ErqAdtq,discussion-id=9ccYZ9Xs,comment-id=0004000000000000,timestamp=2019-09-17T07:08:41.334+0000) > > Analytics are flawed at best No, they are not. If you truly think that (outside of LoL and the specific case discussed here), you likely have no background in statistics and/or probability theory (note: not the dumbed-down version which is studied pre-college). Depending on your mathematical expertise, you may (or may not) be able to fully understand all the relevant theory beyond what I have written. > there is so many extra variables that i would never rely on a number on a site to decide if someone is op or not If that number is a weak metric like WinRate, I wouldn't either.
SFHFWill (NA)
: Not even remotely true. Gnar absorbed a full rotation of spells from 3 people before he backed out. How many rotations should a front line absorb on one champ? We had that meta before, where tanks could easily absorb two or three rotations and it was boring as hell to watch and play. People complained about tanks being OP b/c they couldn't die and pro games were frontline vs frontline with ADC's building tank busting items and just standing behind someone like Sion. I viewed the game completely differently than you. Yeah picks happened and teams backed out. Other times there weren't any picks and team fights were very back and forth.
> [{quoted}](name=SFHFWill,realm=NA,application-id=3ErqAdtq,discussion-id=flXkdui7,comment-id=000700000000,timestamp=2019-10-06T13:35:41.224+0000) > > Not even remotely true. Gnar absorbed a full rotation of spells from 3 people before he backed out. How many rotations should a front line absorb on one champ? Are we talking about the same game? Gnar, in Mega Gnar form, which is _one of the tankiest thing in the game_, **just barely** survived a rotation of two (2) champions: GP and Syndra. At around 36:30, Gnar absorbed: 1. From Nautilus: R (and no more); 2. From Syndra: 2 Q, 1 E and 1 R (Gnar stopwatched Syndra W); 3. From GP: 2 barrels, 1 or 2 autos (it's a bit hard to see). Had Gnar actually absorbed a full-rotation from Nautilus, or even the first tick of GP Ult (which didn't even scratch him), he would have died. In fact, I'm actually most impressed that Gnar didn't die from that, given the meta we are in. > We had that meta before, where tanks could easily absorb two or three rotations and it was boring as hell to watch and play. People complained about tanks being OP b/c they couldn't die and pro games were frontline vs frontline with ADC's building tank busting items and just standing behind someone like Sion. People actually complained, first and foremost, that tanks had too much damage _and_ were also very hard to kill. > I viewed the game completely differently than you. Yeah picks happened and teams backed out. Other times there weren't any picks and team fights were very back and forth. If you watch teamfights even not much long ago, say SKT vs ROX Tigers (Worlds 2016 Semifinals), you will immediately see the huge difference between teamfight now, where teams never fully commit to anything (unless they are stomping enemy team or have some serious advantage), and teamfight back then, when fully-committing in equal or almost-equal situations could actually be done, and there were a pletora of windows for outplays and turn-arounds.
SFHFWill (NA)
: I completely disagree. I see ADC's constantly changing their position and the best ADC champs are the ones that do that the best. {{champion:81}} {{champion:18}} {{champion:145}} {{champion:498}} Those are considered the best ADC's to play and they all have one thing in common, mobility. There are low mobility comps that will use things like {{champion:110}} {{champion:22}} and they play differently. Watch DFM vs ISG, final game of their double round robin for day 3. Incredibly back and forth, lots of trading, aggressive movements. The problem isn't the game balance, the problem is skill. Pros are so insanely good at CSing and maximizing gold that even without a lot of kills they'll still have the income to basically 1 shot anyone caught out of position by landing all their abilities and maximizing their passives or procs and CC chains. Then you go down to something like silver or iron or bronze and the CS is completely crap, abilities don't hit anything and the only reason people die is because they run under turrets or have a 2 minute long fight around a dragon, which is beating on both teams and doing more DPS than any player in the game. The skill gap between pros and the average player that it seems like a 'burst 1 person and GG' at pro play because of vision and map movements that were set up minutes ago, to average play of face checking bush and getting destroyed by 4 people sitting in the bush; is huge. There was a post last year I think by Rito that stated that the average CS above a certain ELO has gone WAY up, leading to 1 shots and snowballing. In short, everyone's skill has risen a fair margin and more people know how to maximize their damage in pro play.
> [{quoted}](name=SFHFWill,realm=NA,application-id=3ErqAdtq,discussion-id=flXkdui7,comment-id=0007,timestamp=2019-10-05T07:35:51.475+0000) > > I completely disagree. [...] Watch DFM vs ISG, final game of their double round robin for day 3. Incredibly back and forth, lots of trading, aggressive movements. That game? Are you sure? That game just proves the main point made by OP: > Right now, the tempo of pro team fights look really similar to the clown fiestas in solo q. Just run around until you can instantly blow one enemy up, then get into a mad scramble to win a 4v5. Almost every teamfight/skirmish played out that way. Here is a (non-exhaustive) breakdown: 14:25 - Skarner (DFM) Flash+R on a flashless Nautilus (ISG), under full vision for Team Blue (ISG). Nautilus gets deleted during suppression, the fight is over. 17:20 - Skarner gets deleted almost instantly by (ISG) Nautilus + Xayah (Gragas last hits). Gnar and Rakan (DFM) try to salvage, but Gnar gets killed in one of the few relatively "slow" deaths of the game, taking around 5-6 seconds to die (in Mega Gnar form). No one dies from ISG, other members of DFM can't do anything. 27:30 - The only teamfight which somehow looks like a teamfight from previous seasons, but only due to the fact that Gragas + Naut couldn't kill Gnar (Xayah wasn't in range and ISG Syndra got suppressed by flash+R of Skarner and then killed), which got a chance to at least use Ult. Gnar still dies, by the way. 28:55 - Nautilus gets Skarner, but then Skarner gets Naut. Guess what? Both get deleted. 36:30 - Xayah gets caught. Gnar, which acted as a frontline for a while (mind you: with the hyper-fed Xayah being dead), just barely escapes death from Syndra R. Gnar also used stopwatch. Starting to see a pattern? Yeah, you get caught, you instantly get deleted and there is nothing your team can do to turn the tables. You frontline? You either barely escape with few hp, or just get deleted for having attempted to actually do your job, i.e., tank for _at most_ 5-6 seconds, most likely this value is bloated due to either stopwatch, skills which make you untargetable, maneuvering time for the carries, etc. Now, the _game itself_ was back-and-forth, sure, that's a fact. The game lasted long, due to the hyper-fed Xayah getting caught, thus artificially prolonging the game. Give the same scenario to a korean team and the game would have ended much, much quicker. Let us not forget that Korean teams closed out games at around 25 minutes during the seasons considered "boring to watch", where the average pro-game (for other regions) was 5-10 minutes higher. But the teamfights? They were fast, catch-based and mostly one-sided (outside of the rare outplay).
Kai Guy (NA)
: > no DuoQ allowed! dont cheat! I wont. So your picks for me to use are Lulu and Yuumi? > Yes, why not. Does that mean anything regarding actual game-impact ? IMHO, no. Not even remotely I feel consistently showing ability to climb off a champions kit is pretty solid evidence for game impact. low/No impact should show up as an inability to climb.
> [{quoted}](name=Kai Guy,realm=NA,application-id=3ErqAdtq,discussion-id=5O8HxP41,comment-id=00060000,timestamp=2019-10-03T20:55:12.816+0000) > > I wont. > > So your picks for me to use are Lulu and Yuumi? Just Lulu, since you already did Yuumi. > I feel consistently showing ability to climb off a champions kit is pretty solid evidence for game impact. low/No impact should show up as an inability to climb. Winning shows skill difference (including macro-game knowledge, mechanics, etc.). IMHO, game impact is an entirely different thing. It has nothing to do with winning/losing. Let me explain with a [short video](https://www.youtube.com/watch?v=Ho9Oe91KzBM) (note that this is an old video, but it's just an example and nothing more). In that video, {{champion:555}} does a penta during a teamfight, winning it. If you could watch the whole game, that moment is certainly the moment you could do a snapshot and say "That's when Pyke had game impact". Red team (Pyke's team) could well lose the game in the next minutes, but the game impact Pyke had is undeniable. So, what actually is game impact? When you watch a game, it's easy to tell which champion had game impact, even if they ultimately lost the game. Defining it in an abstract way is just cumbersome. I believe it has happened at least once to all of us to find a team with a guy who basically did nothing and got carried by the team. We can say he had no game impact (or, at worst, a _negative_ game impact). It is likely we also found the "reverse" situation of a team member who single-handedly carried the game (great game impact). Almost paradoxically, game impact is best shown when your team is losing (not necessarily hard losing, slightly behind is enough) and then there is some "big play" which flips the tables. I think this is the kind of game impact which champions like Lulu will never achieve, while champions like Blitz, Nautilus, Thresh, Pyke, can do much, much more easily. And precisely this fact, for me, is evidence for which champions actually have the kits with the best game impact.
: Coming from someone who mained support from season 4-5ish onwards, I can say with absolute certainty that in order to climb as a support you either need a reliable duo partner with a similar level of skill, or you need to be 3x better than everyone. You can climb with anyone in any role, sure, but some roles require more skill than others to climb with. Like if you play a carry jungler and are decent, you can climb much faster than if you play a mean lulu support. Your lulu support better be the literal meanest. As in absolutely devilish. Support is unique in the sense that you both have the most control over the game's outcome, and the least. You have your hand dipped lightly into every aspect of the game- damage, heals, tanking, engage, disengage, vision, utility, etc., but most of the time you don't necessarily excel immensely in any single one of those aspects or, if you do, it's not enough to carry a game on its own merit. You can enhance others' plays incredibly well, but it's more like you can make good players great and make up for maybe 2-3 out of 5 mistakes bad players make, you don't carry the entire game on your shoulders, unless you are a master of using pyke support or choose a mage support like lux, annie, or brand who are basically carries that sort of support kinda-ish. It's also a thankless job, and if you mess up only a few times, you'll potentially tilt half or more of your team. Like messing up a bard ult or janna ult, or missing a single tornado once as janna and letting your adc die to the khazix early. I have personally given up trying to play support, and I'm going back to top lane. At least there I can have more sway over whether we win or not, and if I mess up a few times and miss one or two spells nobody will be on my case like an adc would in bot lane. If you're determined enough to try this, then I recommend playing heavy engage supports like leona, thresh, or blitz, or heavy disengage supports like janna or nami. Those typically help enhance teamfights and sway them in your favor more than other supports at the moment (except for yuumi, she's busted).
> [{quoted}](name=OfficerKilljoy,realm=NA,application-id=3ErqAdtq,discussion-id=5O8HxP41,comment-id=0007,timestamp=2019-10-04T05:23:24.010+0000) > > Coming from someone who mained support from season 4-5ish onwards, I can say with absolute certainty that in order to climb as a support you either need a reliable duo partner with a similar level of skill, or you need to be 3x better than everyone [...] I didn't quote your whole post only for brevity, but I fully agree with this sentiment.
Kai Guy (NA)
: Can supports climb lower Elo S9? Time to test it.
Very interesting topic. Some supports are more team-dependant than others, that's a fact. So what matters is not if, say, a Diamond smurf can go back in Diamond with his main Nami 1 milion mastery points. That's already a given, or he would have dropped out of Diamond long ago. A little thought experiment can explain what a truly fair comparison would look like. Pick a support which is currently over-tuned, like {{champion:555}} {{champion:53}} {{champion:111}} . Count how many games you need to go from, say, Silver 4 to Gold 4. Now pick a more team-dependant support, like {{champion:117}} or {{champion:350}} (no DuoQ allowed! dont cheat!). Count the same thing. Imagine to repeat this process many many times, in order to get better estimates. Then answer this question: Which champion _actually_ had more impact, statistically? I think every player who played this game since before Season 6 already knows the answer. So, can you pick the lowest winrate support and "smurf" it out of low elo? Sure! Can you do it consistently? Yes, why not. Does that mean anything regarding _actual game-impact_ ? IMHO, no. Not even remotely.
Bârd (NA)
: Skillshots *can* be reliable, like Spirit Fire, but at that point it's almost purely an upside, since you're getting the versatility of a skillshot without the unreliability. As a result, their impact needs to be relatively low. What I'm talking about with "Smite Gameplay" isn't how literally everything is a skillshot, though I'm not a fan of that. Its that there are a ton of low impact skillshots, meaning it doesn't really feel rewarding.
> [{quoted}](name=Bârd,realm=NA,application-id=3ErqAdtq,discussion-id=tU9grzBH,comment-id=00080001000000000000000000010000,timestamp=2019-10-03T15:51:36.441+0000) > > Skillshots *can* be reliable, like Spirit Fire, but at that point it's almost purely an upside, since you're getting the versatility of a skillshot without the unreliability. Nasus W is AoE, good range and basically undodgeable. That's an extreme. Now, think of a line skillshot which gets blocked by the first enemy champion. The skillshot may be relatively fast, which will make it more reliable, and do consistent damage, feeling rewarding. You can tune: damage to champions, CD, range, width of the skillshot, speed, damage to minions (if any), damage penalty if it goes through minions (if any), and a myriad of other variables to make it rewarding. Versatility is a parameter of a skillshot, not a given. > As a result, their impact needs to be relatively low. > > What I'm talking about with "Smite Gameplay" isn't how literally everything is a skillshot, though I'm not a fan of that. Its that there are a ton of low impact skillshots, meaning it doesn't really feel rewarding. We don't have that and we don't need to convert into that.
Hotarµ (NA)
: > First of all, nice visually presented post, Hotarµ. What follows is my input. > > 1. [The "Pyke problem" started earlier than its release](https://boards.na.leagueoflegends.com/en/c/gameplay-balance/NxwIKZN0-the-pyke-problem-started-earlier-than-his-release). I made a long post explaining what factors were involved, but the bottom line is this. In the botlane meta, there are essentially three types of lane: Poke, Kill, Sustain. The hybrid "Poke-Kill" lane, for several reasons, became the unconditional best lane in the game. Mage (and assassin) supports established their dominance over "conventional" (think: enchanters, but not only enchanters by no means) supports. There were/are no viable assassin supports other than Pyke. Maybe in low elos, but overall Pyke was the first of his kind so right off the bat he doesn't fit into the conventional "triangle" of support classes. You're heavily underestimating the strength of "actual" supports because Nami, Soraka, Sona, and Janna have been consistently strong since release and to this day, most if not all of them maintain higher winrates than the "fake" supports (Brand, Zyra, etc.) > 2. Pyke winrate is immaterial, since that metric is affected by high banrate and relatively high (but oscillating) popularity. Such a metric needs to be first manipulated ("normalized") to give it any meaning. For example, banrate needs to be taken into account, while people with little experience on Pyke need to be filtered out. While banrates are important, it is possible to discuss a champion without mentioning them (I did mention banrates in an earlier comment, however.) If his winrate and playrate are already great while maintaining a 40% banrate, that means he would be even _more_ strong (statistically) if his banrate was lowered. By nerfing him in healthy ways and slowly, not all at once, his banrate/playrate/winrate will all start to drop down and better conclusions can be made. And to add to that: Pyke is one of the most hated champions in the game and will be consistently banned regardless of how strong he is. Same goes for Zoe/Akali/Yasuo/Irelia/etc. > 3. Pyke kit is too overloaded and some mechanics need to be removed. There is just no way out of this. You can tweak the numbers all you want, but you aren't gonna get anywhere if he still has all the tools in his kit. There is a way out of it, though. And just for clarity: There is a massive difference between "overloaded" and "overly strong." Pyke has no allied shields, no allied heals, telegraphed abilities, only deals physical damage, can't build HP/MR/armor, and an entirely dodgeble kit. By definition, he is not overloaded. By lessening the strength of Pyke's already present tools, he becomes a weaker and more manageable champion. Adding something like the Grey Health change I suggested makes poke comps more viable into Pyke and reduces his ability to snowball at the drop of a hat. > There is no easy way out for Pyke. Of course, realistically speaking, Riot will barely scratch him with some placebo nerfs... or go over-board and gut him. If the meta changes back to a less damage/kill-oriented meta, indirectly nerfing assassin-like champions, this will also influence Pyke, which may be the "least effort" nerf he could hope for. But there is no guarantee that the meta will evolve towards a generally-lower damage across the board. Pyke's strength is due in large part to the meta, I won't deny that. However, again, there is a solution that keeps his kit similar to what it is now. By nerfing the tools he _has_ and not taking any away, it is possible to balance Pyke.
Thank you for the detailed answer. And sorry for the Wall of Text, but you raised several points. > [{quoted}](name=Hotarµ,realm=NA,application-id=3ErqAdtq,discussion-id=zRfmuAtu,comment-id=000f0000,timestamp=2019-10-03T13:40:27.982+0000) > > There were/are no viable assassin supports other than Pyke. Maybe in low elos, but overall Pyke was the first of his kind so right off the bat he doesn't fit into the conventional "triangle" of support classes. I typed "assassin" only to specifically include him (I didn't want to sponsor playing {{champion:7}} support, sorry if it sounded that way). However, Pyke still fits in the support triangle (albeit unconventionally) under the tag: Kill Lanes. > You're heavily underestimating the strength of "actual" supports because Nami, Soraka, Sona, and Janna have been consistently strong since release and to this day, most if not all of them maintain higher winrates than the "fake" supports (Brand, Zyra, etc.) Honestly, I would really like to heavily under-estimate their strength, but the sad truth is that (with one or perhaps two exceptions) they are just straight-up outclassed. I have played many games (50+ in most cases) with all the champions below and watched many people play them in Diamond+. Here are my conclusions: {{champion:37}} : Trash-tier champion. Barely considered a champion for the longest time. The last time I saw Sona in pro-play was due to Sona+Taric abuse. {{champion:117}} : B-tier support. Barely that. I mained Lulu in S5, when she was literally _the_ definition of "flex champion" and quite meta-warping (in pro-play). That epoch has long ended. Riot avoided the "generalist" problem Lulu has by gutting her. Now, almost every single support in the game can do better than Lulu. {{champion:267}} : Nami may be an A-tier support, but can become Trash-tier also quite easily. A good Nami will get systematically countered by a good Brand, only skill-difference can make up for this skewed matchup. And the fact is that Nami has problems against {{champion:63}} as well as many other supports. {{champion:16}} : A very special case. Soraka, considered by her strength/weakness alone, is B-tier. If a sufficiently broken synergy on heals exist in the game, she can skyrocket to A+ tier (or even above). Soraka depends **heavily** on how much the game can break when "Healing" is optimized. Numbers matter a lot here. {{champion:40}} : S-tier, almost meta-indifferent champion. _**Janna is the exception, not the rule**_. She brings huge disengage potential, which is just too damn good utility in any meta this game has ever had. Janna will always be up there in the "holy mountain" of supports. As a whole, they are outclassed by the likes of: {{champion:555}} , {{champion:412}} , {{champion:12}} , {{champion:497}} , and currently also {{champion:111}} , {{champion:53}} . Blitz and Naut may be transitory S-tier, sure, but Thresh is always gonna be **the** meta-indifferent, permanently S-tier, support. Rakan is always gonna be the super-engage, S-tier support (also meta-indifferent, unless Xayah and all other ADCs which can synergize with Rakan are gutted into oblivion). Pyke is on a whole other level entirely (meta-warping, i.e. S+). You may object that enchanters winrates are generally fine. It would require me a separate thread to explain why enchanters were made over-tuned (number-wise) and why their winrates are almost entirely irrelevant. And also why some nerfs were completely undeserved. Aside from the Ardent Censer meta (everything taken to extremes becomes horrible), enchanters are basically the "healthiest" thing we got in the Support Meta. To be really quick: everything else assumed equal, who would give you the best chances to win the game, between: (the gold-equivalent of) a 10/0 Pyke and (the gold-equivalent of) a 10/0 Nami? We all know that a hyper-fed assassin just takes over games (if he is good, skill-wise). A hyper-fed enchanter, alone, cannot (care: this is true in SoloQ, not in competitive). Still, I would take the 10/0 Nami any day of the week. > While banrates are important, it is possible to discuss a champion without mentioning them (I did mention banrates in an earlier comment, however.) > > If his winrate and playrate are already great while maintaining a 40% banrate, that means he would be even _more_ strong (statistically) if his banrate was lowered.By nerfing him in healthy ways and slowly, not all at once, his banrate/playrate/winrate will all start to drop down and better conclusions can be made. Perfectly agree. > And to add to that: Pyke is one of the most hated champions in the game and will be consistently banned regardless of how strong he is. Same goes for Zoe/Akali/Yasuo/Irelia/etc. I don't hate him and feel a bit bad for people hating Pyke, since the champion itself is only the outcome of a series of deeper flaws. But what you stated are facts. Pyke banrate will stay high, that's a given. > There is a way out of it, though. > > And just for clarity: There is a massive difference between "overloaded" and "overly strong." Pyke has no allied shields, no allied heals, telegraphed abilities, only deals physical damage, can't build HP/MR/armor, and an entirely dodgeble kit. By definition, he is not overloaded. Which definition are you using? On the contrary, _Pyke is probably the most overloaded support!_ There is a guy who compiled a spreadsheet of almost all champions in LoL. The most overloaded one was Camille, but Pyke was up there. There is just too much in his kit: stealth, grey health, execute, R reset, slow, grab (telegraphed or not, it's a grab), good AD damage (which is by no means a given for a support), through-walls mobility, AoE stun, (pre-nerf) AoE damage, unique gold-sharing mechanic, movement speed boost. I probably missed something. He doesn't have anything of the classical support kits, i.e. heals and shields, but only because he doesn't need them to function as a champion. As you said, strength doesn't have anything to do with overloading. Camille isn't OP right now, clearly. She still is overloaded as hell, likely the most overloaded champion in the game. Followed by Ekko, iirc. > By lessening the strength of Pyke's already present tools, he becomes a weaker and more manageable champion. Adding something like the Grey Health change I suggested makes poke comps more viable into Pyke and reduces his ability to snowball at the drop of a hat. So, you are suggesting Poke lanes, which have classically been dumpstered by Grab Lanes since well before this meta arrived, to become Pyke counter? I cannot imagine that, unless you remove Pyke grab. I mean, Lux vs Pyke (both supports) will be basically a one-sided win for Pyke, grey-health or not. Same for Brand vs Pyke. > Pyke's strength is due in large part to the meta, I won't deny that. That's the first thing which needs to be changed. > However, again, there is a solution that keeps his kit similar to what it is now. By nerfing the tools he _has_ and not taking any away, it is possible to balance Pyke. Pyke stealth can be removed and Pyke can still be Pyke, without losing identity. Yes, stealth is too much. How many champions in the entire roster have stealth and how much do they pay, power-budget wise, for it? Very few have it, only one, i.e. {{champion:28}} , gets it "for free". This needs to go together with the Grey Health nerf you proposed. Though, it's highly unlikely that Riot will touch Pyke again, unless he dominates in pro-play at Worlds. Looking at the power-budget, Pyke just has too much for a champion. Compare him to old champions (like Annie, Garen, Tristana) and you will see how many tools Pyke has in his kit.
Hotarµ (NA)
: Pyke: Yes, he needs to be nerfed. A 560k mastery point main's perspective.
First of all, nice visually presented post, Hotarµ. What follows is my input. 1. [The "Pyke problem" started earlier than its release](https://boards.na.leagueoflegends.com/en/c/gameplay-balance/NxwIKZN0-the-pyke-problem-started-earlier-than-his-release). I made a long post explaining what factors were involved, but the bottom line is this. In the botlane meta, there are essentially three types of lane: Poke, Kill, Sustain. The hybrid "Poke-Kill" lane, for several reasons, became the unconditional best lane in the game. Mage (and assassin) supports established their dominance over "conventional" (think: enchanters, but not only enchanters by no means) supports. 2. Pyke winrate is immaterial, since that metric is affected by high banrate and relatively high (but oscillating) popularity. Such a metric needs to be first manipulated ("normalized") to give it any meaning. For example, banrate needs to be taken into account, while people with little experience on Pyke need to be filtered out. 3. Pyke kit is too overloaded and some mechanics need to be removed. There is just no way out of this. You can tweak the numbers all you want, but you aren't gonna get anywhere if he still has all the tools in his kit. There is no easy way out for Pyke. Of course, realistically speaking, Riot will barely scratch him with some placebo nerfs... or go over-board and gut him. If the meta changes back to a less damage/kill-oriented meta, indirectly nerfing assassin-like champions, this will also influence Pyke, which may be the "least effort" nerf he could hope for. But there is no guarantee that the meta will evolve towards a generally-lower damage across the board.
Bârd (NA)
: 1. Yes, we can go back to "low but reliable damage", it's just that you're gonna piss off the Rengar and LB players to the point they quit (oh no... don't go...) 2. No, point and click is definitely the way to go if you're going to have low damage. If you make everything a low damage skillshot, the result is Smite. Point and Click does not equate to no skill, it simply means that mechanical skill isn't the kind of skill that matters most. If you have a game with a handful of targeted disables and an ocean of powerful skillshot nukes, of course they're going to be broken, you've balanced the game around dodging everything. You simply need to balance around the idea of "Ok I'm going to get stunned here, what do I do afterwards?" instead of "Ok I got stunned now I'm just dead". 3. Yes, they exist. You can simply rebalance them around a longer damage window. Again, it's going to piss off Rengar and LB mains, but fuck em. 4. Under no circumstances should the game remain as-is.
> [{quoted}](name=Bârd,realm=NA,application-id=3ErqAdtq,discussion-id=tU9grzBH,comment-id=000800010000000000000000,timestamp=2019-10-02T21:22:19.595+0000) > > 1. Yes, we can go back to "low but reliable damage", it's just that you're gonna piss off the Rengar and LB players to the point they quit (oh no... don't go...) > 2. No, point and click is definitely the way to go if you're going to have low damage. If you make everything a low damage skillshot, the result is Smite. Point and Click does not equate to no skill, it simply means that mechanical skill isn't the kind of skill that matters most. If you have a game with a handful of targeted disables and an ocean of powerful skillshot nukes, of course they're going to be broken, you've balanced the game around dodging everything. You simply need to balance around the idea of "Ok I'm going to get stunned here, what do I do afterwards?" instead of "Ok I got stunned now I'm just dead". > 3. Yes, they exist. You can simply rebalance them around a longer damage window. Again, it's going to piss off Rengar and LB mains, but fuck em. > 4. Under no circumstances should the game remain as-is. 1. I am (to be precise, was) a {{champion:7}} player and you are not gonna piss me off. LB has other problems anyway. Believe it or not, a balanced game is better for everyone, including the ones whose champions get nerfed (which may seem like a paradox, but it is not), though a vocal minority may whine for some period of time. 2. Simply put, you are wrong. Skillshots can be reliable too. The long answer is in my first (chronological order) post in this topic, which you should read quite carefully. The result isn't Smite because in Smite literally everything, including AAs, are skillshots, which will never happen in League. 3. No further comment is needed here. 4. Agreed.
Bârd (NA)
: Ditto. Leblanc is a terrible design through and through. Qiyana is a decent design, but absolutely should not be balanced as an assassin. She should be the next Irelia, not the next Talon.
> [{quoted}](name=Bârd,realm=NA,application-id=3ErqAdtq,discussion-id=tU9grzBH,comment-id=0008000100000000,timestamp=2019-10-02T19:54:18.066+0000) > > Ditto. > Leblanc is a terrible design through and through. > Qiyana is a decent design, but absolutely should not be balanced as an assassin. She should be the next Irelia, not the next Talon. So, apparently, we can't go back to your point one (low but reliable damage, with the tweaking that "reliable" doesn't mean point-and-click but actually requires a modicum of skill, and "low" is more like "medium-low" or "conditionally medium-high"), but since you admit the existence of a whole set of champions with high damage and high reliability, we should just ... what? Keep it as it is, without giving voice to the blatantly obvious and obnoxious "damage meta" ? We can't have both ways, can we at least have the decent one and scrap the current one? Apparently, the answer is no. If so, the topic should be titled "Guys, you can't have it any way".
Bârd (NA)
: A mistake that it's too late for the devs to go back on.
> [{quoted}](name=Bârd,realm=NA,application-id=3ErqAdtq,discussion-id=tU9grzBH,comment-id=00080001,timestamp=2019-10-02T19:39:43.190+0000) > > A mistake that it's too late for the devs to go back on. What about {{champion:7}} ? Or, using a modern example, {{champion:246}} ?
Bârd (NA)
: Guys, you can't have both ways.
This thread raises some interesting points and, at first glance, may appear as a "quality" post. Let's see if it holds up to a reality check (spoiler: it doesn't). > When it comes to Mobas, there are basically three design philosophies. > > 1. Damage is relatively low, but reliable. As a result ,the game is more oriented towards tactics than mechanics (Dota 2, old League). > 2. Damage is high but unreliable. As a result, the game is heavily oriented towards flashy mechanical outplays (Current League). > 3. Damage is relatively low *and* unreliable. As a result, your game is too mechanically oriented for the Tactical players and isn't flashy enough for the mechanical players, so *nobody plays it* (Smite, Heroes of the Storm). You are missing: 4. Damage is high and reliable (also current League). But before that, we need to take a step back and realize that: 1. _**Damage is not a binary variable**_ (high/low). Damage is a _number_, so it can vary from 0 to relatively high values. It is often useful to "normalize" (sorry for the technical term) damage between 0 and 1 (with 0 meaning literally zero, while 1 basically means either the theoretical max damage, which may be infinite, or some other very high, but finite, number, which is typically more useful when you have infinitely-scaling champions). 2. _**Reliability is not a binary variable**_. Reliability is actually (surprise?) another _number_, which varies between 0 and 1. Some of you may have guessed (correctly) that reliability is a probability (that depends on several factors), but let's keep this simple for the non-mathematical adepts. Thus, when designing a champion, instead of having only the 3/4 possibilities listed above, the two "degrees of freedom" (damage, reliability), or better, (normalized dmg, reliability), generate all the couples between (0,0) and (1,1), i.e., the unit square. Any point in the unit square gives you a potential champion. In principle, all combinations are possible. The three design philosophies listed above are just an arbitrary simplification (made by the original poster), but in reality there is no such restriction. So, you can have, say, moderate damage with moderate reliability (which doesn't fit in any category listed by the original poster). > [{quoted}](name=Bârd,realm=NA,application-id=3ErqAdtq,discussion-id=tU9grzBH,comment-id=,timestamp=2019-10-02T02:30:52.467+0000) > > If Riot releases a champion full of skillshots, they are going to have a ton of damage to make up for that damage being unreliable. Not necessarily. > Yes, damage is out of control right now. Agreed.
: > [{quoted}](name=Xintium,realm=EUW,application-id=3ErqAdtq,discussion-id=KpjLsWb5,comment-id=000c,timestamp=2019-09-27T10:01:30.560+0000) > > Watch pro-play (before Pyke nerfs), check banrates in high elo, and you will see your argument crumble. Or maybe it is just the ADC's I play. Interrupt any part of his combo and he loses almost all effectiveness. {{champion:15}} {{champion:18}} {{champion:81}} and {{champion:51}} all have tools in their kit to interrupt his combo. Some before the hook others when he goes to stun. Some of these tricks I learned watching pro-play. (Sivir is self explanatory). I don't play them but I'm pretty sure {{champion:145}} and {{champion:498}} could interrupt his shit too.
> [{quoted}](name=NaughtyWord,realm=NA,application-id=3ErqAdtq,discussion-id=KpjLsWb5,comment-id=000c0000,timestamp=2019-09-27T11:02:21.518+0000) > > Or maybe it is just the ADC's I play. > > Interrupt any part of his combo and he loses almost all effectiveness. Note that Pyke doesn't necessarily want to go for the ADC. It heavily depends on the matchup. Anyway, if you interrupt any part of his combo without dealing damage to him ("damage" here means something that sticks, i.e., his grey health won't regen that), he still somewhat wins the "trade", since his ADC could punish you for not having your escape. And Pyke threat is still online, just not position-wise, but damage-wise Pyke is almost always online. > {{champion:15}} {{champion:18}} {{champion:81}} and {{champion:51}} all have tools in their kit to interrupt his combo. Some before the hook others when he goes to stun. Some of these tricks I learned watching pro-play. (Sivir is self explanatory). I perfectly agree with Sivir and Ezreal: they counter all hook supports, they are not a Pyke-specific hard-counters. Caitlyn has safety due to her range and E. Tristana is tricky: Tristana W at rank 1 has 22 seconds CD, while Pyke has almost half that CD with his Q. Pre-6 Tristana needs very careful positioning against Pyke. > I don't play them but I'm pretty sure {{champion:145}} and {{champion:498}} could interrupt his shit too. Xayah pre-6, absolutely not. She needs to rely: (1) on a support (preferably Rakan) to keep her alive; (2) her numbers being over-tuned. Kai'Sa is just being Kai'Sa and can theoretically counter anything. In practice, however, Kai'Sa pre-6 is still vulnerable. Finally, Pyke truly shines during teamfights via its R resets: that's his strongest point, specifically what won the game for G2 vs SKT at MSI. Pyke is not a "true" lane-dominant champion and should not be judged based on that. In fact, he often wasn't played support in pro-play. It's his scary teamfight instant cleanup power. He still needs a team behind him to do the first line of damage, of course, but after a successful teamfight, Pyke just makes his whole team snowball out of control. Normally, you would have only one champion getting fed out of a triple kill, but with Pyke the whole enemy team gets fed for free. This mechanic is really snowbally and makes him have a unique strength. In a way, you could say that Pyke is just an amplifier. But, in this meta specifically, an amplifier of damage and an amplifier of snowball is really scary. Regardless of all that, Pyke dominated pro-play for sufficiently long to show that he was indeed a problem.
: Let's Talk Hook Supports
First of all, very nice thread, **DivineMantra**. I feel a bit bad for you, since some uneducated kids still don't know how to engage in a civil discussion, expressing their opinion (pro or contra), without recurring to laugh, ad-hominem attacks and vulgarity. I am a support main who played almost every single champion in the game as support (not in ranked, unless it was a niche pick). I played Brand and Lux support well-before they were even remotely considered supports. I started the game with Sona/Janna/Lulu, but I played many games with Braum, Annie (when Annie support was a pro-pick), Karma and Zyra, alongside Leona and Nautilus. But my favorite support will always be Morgana. Finally, while not being even remotely good at this game, I avidly watched pro-play (until now, S9 was really bad to watch, but still watched a good part of it). These are my thoughts: {{champion:89}}: She is balanced and I don't think she fits into this discussion. {{champion:412}}: The ever-green, permanently in-meta, support. The only reason this champion doesn't have stellar winrate is that Thresh is very hard to master and mechanics-intense (and Elo matters much less than one may think). I have seen a Diamond dedicated Thresh main who single-handlely "smurfed" in Diamond and won botlane super-hard. The same guy wasn't so stellar in the following game. Thresh is _insanely_ difficult to play at a very high level. Sometimes even pro-players screw-up with him! Being "average" with Thresh is just common and an average Thresh is actually a "bad" Thresh, compared to what the champion can do when piloted correctly. Consistently S+ Thresh are basically Challengers and pro-players. Having said that, WR is immaterial when judging Thresh, because Thresh has very high popularity: only a fraction of that are dedicated Thresh mains. I highly doubt Thresh will ever be significantly nerfed by Riot and he will stay in the meta for as long as the game survives. In short, Thresh is the perfect example of a champion who will be permanently skill-capped by the human piloting it. {{champion:53}}: Buffing Blitz Q range is a mistake. The hitbox is already a mess. Fix that first, then we can talk. Period. {{champion:111}}: Nautilus, as support, should only exist as a counter-pick to Blitz (and perhaps few other supports). At least, so the matchup was back in the day. There was a time when one Naut combo, Q+auto+E (+W to avoid retaliation damage) halved the ADC hp bar. I am not familiar with what happens now as a % (can someone tell me? including aftershock dmg). The hitbox is huge, the hook is relatively easy to land, the point-and-click R is just too good for a 2v2 skirmish in botlane, the E damage+slow was insane and I don't remember if they nerfed it (they probably did long ago). Aftershock is disgustingly OP on him. Still, it seems time to have a deep look at him. Definitely on the nerf side. {{champion:555}}: I don't hate Pyke. He was the outcome of a long chain of things (mostly bad things) that Riot did. But, holy moly, isn't Pyke _the most over-loaded support ever!_ (excluding the short-period where Camille support was a thing). Play this champion robotically perfect and you can win any game at any elo, including pro-play, having fun trashing literally everyone with your "skillz". No, this isn't true for all champions. Not even for 50% of the roster, thank you for asking. Pyke WR is simply limited by its insanely high banrate and also Pyke has a high skill ceiling. Again, WR is a corrupted metric here. If someone will ever make an Artificial Intelligence (AI) for LoL, like they did Open AI for Dota2, please, make an AI for Pyke, just to show everyone how utterly, insanely broken and meta-warping this champion is, _despite the nerfs_ he already took in the face and laughed off. Monster utility, monster multi-target CC, stealth, movement speed buff, unique gold-sharing mechanic, borderline-perfect execute, good (if somewhat telegraphed) hook, nice Q damage, insane mobility, grey-health, ... I mean, what more do you want? And I didn't list everything. And this comes from a guy who enjoyed playing Pyke when he was released. With all the non-numeric power his kit has (basically the combined power-budget of 2-3 old champions, like Annie), either his numbers get gutted or Pyke is gonna rise up to where Thresh currently is: the ever-green, meta-indifferent, support holy mountain. He was certainly at the apex of the support food chain before the nerfs, now he probably shares his undisputed first place with others. Despite all of that, I still don't hate Pyke. But he just has too much in his kit. Something must be removed to make room for balance. Here, "something" does not mean numbers, that approach is just Riot delusion that tweaking numbers can balance everything, **_mechanics must be touched, most likely removed or significantly changed_**. For now, just keep banning him. > [{quoted}](name=NaughtyWord,realm=NA,application-id=3ErqAdtq,discussion-id=KpjLsWb5,comment-id=000b,timestamp=2019-09-27T08:04:14.869+0000) > > I don't even think Pyke is all that great. He's very scripted which means its easy to deal with his wambam thank you ma'am combo. Watch pro-play (before Pyke nerfs), check banrates in high elo, and you will see your argument crumble.
HeeroTX (NA)
: > [{quoted}](name=Xintium,realm=EUW,application-id=3ErqAdtq,discussion-id=iHeFMl0J,comment-id=000100010001000100000001000000000000,timestamp=2019-09-26T17:55:55.472+0000) > > What you described is basically the best scenarios ever: a champion having significat influence without killing enemies himself. If players whine in that situation, let them whine, cause they don't understand basics. However, I highly doubt that players consistently whined about that There's a thread RIGHT NOW complaining about "hook supports" ALL hook supports. Pyke sure, Blitz with the buff, I get, but also Thresh, Naut & Leona. These are champs whose kit is literally designed to SETUP plays, and yeah, when played well, they're annoying as hell, but they provide utility rather than damage. Remember when Soraka was strong? Healing is a pain in the ass to deal with, so it got nerf. Remember when Tahm was one of the best supports, back when he wasn't a top laner? And on and on. But ignore that, you know how many games I see every week where people bitch about the OP midlaner or bot without noticing the 50 assist support? Especially when it's a "support" Pyke, or Brand or Velkoz complaining about the team not matching up to their 8 kill skill level? Or all the "tanks" that build AP and wait BEHIND the mid and bot so they can "LC$-Big-plays" assassinate someone? Hell, the most annoying support right now (IMO) is Lux, simply because she can spam damage AND utility just mashing buttons to throw out everything in her kit on insane cooldowns. Yeah, the pros play all sorts of things because they coordinate well. But you've got champs like Kogmaw who are super strong but see minimal play because 9 times out of 10 (in solo-q) no one is gonna cover for you and you're screwed with no mobility. Note, the above does NOT apply to "high elo", this is Gold and below, but that population is the vast majority of the League playerbase. And again, THAT is also why we "need" autofill, because people won't play the "non-damage" roles, and if those roles have their PROPER influence, then you get a ton of complaining about bad games because the team with 1 utility player beats the team with 0 most of the time and all the people that want to play heavy kill supports (which is most of them) whine and bitch and complain about how they never get to play mid.
> [{quoted}](name=HeeroTX,realm=NA,application-id=3ErqAdtq,discussion-id=iHeFMl0J,comment-id=0001000100010001000000010000000000000000,timestamp=2019-09-26T18:30:43.479+0000) > > There's a thread RIGHT NOW complaining about "hook supports" ALL hook supports. Pyke sure, Blitz with the buff, I get, but also Thresh, Naut & Leona. These are champs whose kit is literally designed to SETUP plays, and yeah, when played well, they're annoying as hell, but they provide utility rather than damage. I have read that thread. You had a really wrong impression of it, since the original poster (a Diamond support player) even admits to be biased himself. He plays some enchanters who get systematically hard-countered by Thresh, Blitz, Naut, Pyke. Leona is the only true exception there, but he himself says she is balanced. He raises some specific points and I agree with him for most of them. > Remember when Soraka was strong? Healing is a pain in the ass to deal with, so it got nerf. Remember when Tahm was one of the best supports, back when he wasn't a top laner? And on and on. The nerf Soraka got was, numbers-wise, badly tuned, but at least in theory justified. Tahm was a bit too much, honestly, and his ult was simply too good in pro-play, while being barely viable in soloQ (due to lack of coordination). > But ignore that, you know how many games I see every week where people bitch about the OP midlaner or bot without noticing the 50 assist support? Especially when it's a "support" Pyke, or Brand or Velkoz complaining about the team not matching up to their 8 kill skill level? Or all the "tanks" that build AP and wait BEHIND the mid and bot so they can "LC$-Big-plays" assassinate someone? Let them bitch. Silence is a rarely-apprecciated virtue. > Hell, the most annoying support right now (IMO) is Lux, simply because she can spam damage AND utility just mashing buttons to throw out everything in her kit on insane cooldowns. Lux support is the outcome of a long chain of effects. She is very annoying, since one Lux full-combo (Q+E+R + aa + passive procs + ignite) bursts really hard and can 100-0 the ADC (even with {{summoner:7}}, because that gets countered by {{summoner:14}} ) if Lux has {{item:3802}}. However, Lux rarely gets to use that combo, it requires a big mistake from the enemy; most of the time is just Q+E+R, which is strong but you can survive that, at least during laning phase (in late game, not so much). Lux weakness, however, is her total lack of mobility (she doesn't even have a speed boost, like Orianna W and now Annie E). Lux support isn't strong and never will be, there are just too many strictly better supports, like Pyke, who basically does the same thing (kill you with a full-combo), but: (1) it doesn't require a big mistake from the enemy; (2) it is significantly easier to land; (3) Pyke has disgusting mobility and safety; (4) snowballs crazy hard, due to the gold-sharing mechanic. > Yeah, the pros play all sorts of things because they coordinate well. But you've got champs like Kogmaw who are super strong but see minimal play because 9 times out of 10 (in solo-q) no one is gonna cover for you and you're screwed with no mobility. > > Note, the above does NOT apply to "high elo", this is Gold and below, but that population is the vast majority of the League playerbase. And again, THAT is also why we "need" autofill, because people won't play the "non-damage" roles, and if those roles have their PROPER influence, then you get a ton of complaining about bad games because the team with 1 utility player beats the team with 0 most of the time and all the people that want to play heavy kill supports (which is most of them) whine and bitch and complain about how they never get to play mid. Excluding perhaps Diamond+ or even higher elos (and sometimes even there), the team with the pure-utility support, nowadays, is most likely going to lose, not win. That's why no one wants to play that and the transition happened in the first place. Compare Karma/Yuumi/Lulu/Sona winrates with the winrates of Naut/Blitz/Pyke (care with Pyke, he has lower WR merely due to its banrate being insanely high, so you need to adjust for that). Thresh is also up there. Thresh winrate isn't so high due to its very high popularity, while being a very hard champion to master, so few people can play Thresh at a high level, the others are just average or straight-up bad with him (note: elo doesn't matter much, it is incredibly easy to screw-up with Thresh), I can immediately see the difference between a dedicated Thresh main with 200+ games and a Thresh with barely 50 games on him, the mechanics gap is just huge. In this case, the problem is people not capable of piloting Thresh, rather than the champion, which is (and always will be) an ever-green in all metas, permanently in the strong side of the power spectrum.
HeeroTX (NA)
: Since my post apparently gave the wrong impression, let me clarify something. I do not LIKE the current situation nor am I HAPPY about the proliferation of damage. I main Tristana (for various reasons) but after that my PREFERENCE is to play supports and/or tanks. The PROBLEM I'm pointing out is that when you have what I (personally) consider a "healthy" META, then tanks and (true) supports have a LOT of influence on a game, they just don't KILL a lot of enemies. The result of this is that the 95% of players that don't WANT to play those roles then WHINE incessantly because their attempts to carry late game with a top duelist against a well formed team with a solid tank frontline fail because they can't 1v3 split push to victory. Just look at the argument being had. "Tank META" should have never been a problem, because BOTH sides should HAVE tanks. I remember OP nigh-unkillable Mundo. You know what the biggest problem was then? PEOPLE DIDN'T BUILD RIGHT. Tanks were super strong, but you had a bunch of braindead lemmings just building X without modifying for what their opponents had or were doing. The "META" should always include at least one of the following: 1. Tank (usually top) 2. Burst (usually Mid) 3. DoT (usually ADC) 4. Utility (usually support) And then #5 can be whatever is strong atm or what fits well in the comp. The problem is everyone wants to play #2, a decent number are good with #3, and #1 & #4 are usually tough sells. This is also why I've long maintained Riot should just make URF a permanent gamemode, not because I LIKE it, not because I want to play it (I don't) but because then all the people that want to play "Team Deathmatch" can do so and those of us that want a true "strategy" game can have it back. (EDIT: Riot instead decided they want to keep their SR stats high so they turn SR into "high damage" rather than see numbers drop by having URF always on) ADDENDUM: The above is also why autofill is a necessary thing btw.
Thanks for the clarification, it was much needed. > [{quoted}](name=HeeroTX,realm=NA,application-id=3ErqAdtq,discussion-id=iHeFMl0J,comment-id=00010001000100010000000100000000,timestamp=2019-09-26T16:45:11.331+0000) > > The PROBLEM I'm pointing out is that when you have what I (personally) consider a "healthy" META, then tanks and (true) supports have a LOT of influence on a game, they just don't KILL a lot of enemies. The result of this is that the 95% of players that don't WANT to play those roles then WHINE incessantly because their attempts to carry late game with a top duelist against a well formed team with a solid tank frontline fail because they can't 1v3 split push to victory. What you described is basically the best scenarios ever: a champion having significat influence without killing enemies himself. If players whine in that situation, let them whine, cause they don't understand basics. However, I highly doubt that players consistently whined about that, because such a situation never occurred for a significant timeframe in LoL (Ardent Censer is the only real exception, but an over-tuned item is quite different from an entire class of champions being over-tuned). > Just look at the argument being had. "Tank META" should have never been a problem, because BOTH sides should HAVE tanks. I remember OP nigh-unkillable Mundo. You know what the biggest problem was then? PEOPLE DIDN'T BUILD RIGHT. That is, frankly speaking, utterly false. (1) Both sides had tanks, but not all tanks are created equal; (2) In the tank meta, you could easily predict the outcome of a pro-play match by: (2.1) counting tanks (team with most tanks > team with less tanks); (2.2) if they have the same number, the team with the best tanks would typically win. Team-comp mattered only to the bare minimum of not having anti-synergies. People did build right. At least, pro-players did. The problem is: no "right" build existed to counter tanks, unless you were a strictly better tank, in which case, you just trashed the enemy tank. There was no real build to mitigate what tanks could do. > Tanks were super strong, but you had a bunch of braindead lemmings just building X without modifying for what their opponents had or were doing. This happened for Midlaners and ADCs, but only because every single possible combination of items in the game was just trash compared to the raw power of tanks. Builds were stagnant because builds were almost irrelevant. Yes, it was that bad. > The "META" should always include at least one of the following: > 1. Tank (usually top) > 2. Burst (usually Mid) > 3. DoT (usually ADC) > 4. Utility (usually support) > And then #5 can be whatever is strong atm or what fits well in the comp. The problem is everyone wants to play #2, a decent number are good with #3, and #1 & #4 are usually tough sells. This is also why I've long maintained Riot should just make URF a permanent gamemode, not because I LIKE it, not because I want to play it (I don't) but because then all the people that want to play "Team Deathmatch" can do so and those of us that want a true "strategy" game can have it back. > > (EDIT: Riot instead decided they want to keep their SR stats high so they turn SR into "high damage" rather than see numbers drop by having URF always on) What you have written here is essentially true. The sad truth is that we aren't gonna get back to a strategy game in any way... because money. The few dedicated players who wanted to be #4 are now basically "forced" into playing supports like Pyke, or be strictly sub-optimal (which is not a real choice). Reducing utility always implicitly leads to damage-dealers being stronger, since you can't keep them at bay with utility anymore.
HeeroTX (NA)
: "Tanks needed more damage" for one simple reason that is (and unfortunately ALWAYS will be) true: Teamwork OP. The core problem with this game is you have maybe 5% of people that are 100% willing to play whatever role is necessary to help the team win and actually ENJOY working in a team environment. Everyone else just wants to be "big dick daddy". This is why true "tanks" are not a thing, because if you have a viable tank, then a small percentage of players who play "correctly" will dominate and everyone **who doesn't WANT to play "correctly"** will complain about "muh agency". I still believe you could improve this a LITTLE bit if Ranked wasn't solely determined by W/L, but at the end of the day, stats prove that the VAST majority don't want to play roles that aren't killing lots of opponents. That's why tanks are all but useless and why you have so many mages and assassins in the "support" role.
> [{quoted}](name=HeeroTX,realm=NA,application-id=3ErqAdtq,discussion-id=iHeFMl0J,comment-id=000100010001000100000001,timestamp=2019-09-26T14:34:58.004+0000) > > "Tanks needed more damage" for one simple reason that is (and unfortunately ALWAYS will be) true: > Teamwork OP. Can you please explain how "Teamwork OP" somehow makes tanks needing more damage? Aside from regurgitated slogans, I mean. > The core problem with this game is you have maybe 5% of people that are 100% willing to play whatever role is necessary to help the team win and actually ENJOY working in a team environment. Everyone else just wants to be "big dick daddy". This may or may not be true. I was a support main back in the day (playing "true" supports, like Sona, Janna, Lulu, Braum, not "modern" supports like Pyke), so I certainly can't sympathize with this feeling. > This is why true "tanks" are not a thing, because if you have a viable tank, then a small percentage of players who play "correctly" will dominate and everyone **who doesn't WANT to play "correctly"** will complain about "muh agency". True tanks, with very few exceptions, are not a thing now due to the meta being unfavorable to them. By the way, if a tank deals _less_ damage, the damaged player will have _more_ agency, since it will have time to do something. I don't know if you were around when the tank meta was a thing, but, at that time, tank players literally warped and dominated the meta and everyone who didn't play "correctly", i.e., by picking a champion from a very small subset of the champion roster, was left behind. And the problem of that was: (1) not an absence of teamwork; (2) not the "big dick daddy" (those were quickly culled out); but (surprise!) _tanks dealing too much damage!_ > I still believe you could improve this a LITTLE bit if Ranked wasn't solely determined by W/L, but at the end of the day, stats prove that the VAST majority don't want to play roles that aren't killing lots of opponents. That's why tanks are all but useless and why you have so many mages and assassins in the "support" role. We have mages and assassins in the support role due to a _combination of factors_ (I have explain these in more detail in a previous post regarding Pyke), but the short version is: "Kill Lanes" became the best lanes, unconditionally. In the land of "kill before you get killed", assassins and burst mages were, quite trivially, the best choices available. And Pyke, with his gold-sharing mechanic and officially labeled as "assassin-support", sealed the deal. It may be perfectly realistic that the vast majority of players don't want to play "true" supports or, more generally, champions who can't burst another champion. Still, I don't see why this should lead to tanks dealing tons of damage being "ok". It's not the way to fix the problem anyway, it only exacerbates it, by adding tanks to the problem! That's why, as I said, most of the reasons justifying "tank meta" were, under critical scrutiny, nonsense.
: > The "power budget" philosophy is perfectly fine in theory, but I haven't seen any realistic application of it. Also mobility seems to have way too little cost to the power budget compared with how powerful it is since most damage (outside of Auto attacks) are skill shots which mobility helps avoid or helps getting in position to land your own skill shots And it makes sense as the value of mobility depends a lot... as the higher your damage and the lower your durability the more valuable mobility is or if it can be used to kite and outrange enemies not allowing them to fight back (Prime example being Ezeral with his decently long range Q combined with the mobility of his E) But mobility isn't very useful if you can get to your enemy but don't have the CC or damage to lock set up allies or kill the enemy themselves (of course it is still useful for staying alive and all champions do have the potential to deal decent damage just with Auto attacks and enough ADC items)
> [{quoted}](name=Champion Skin,realm=EUW,application-id=3ErqAdtq,discussion-id=yZIvfMTw,comment-id=0000000000000000,timestamp=2019-09-26T12:10:13.571+0000) > > Also mobility seems to have way too little cost to the power budget compared with how powerful it is since most damage (outside of Auto attacks) are skill shots which mobility helps avoid or helps getting in position to land your own skill shots This is true. To the point that mobility is basically mandatory on new champions. Now being immobile (like Annie, excluding the recent change to her E) feels almost like playing a different game. Nothing compared to the 2+ dashes/mov.-speed buffs/etc. that new champions get essentially for free.
Eedat (NA)
: > Unfortunately, newer champions basically get (A,B) for free, can do C with some effort and can also do that extra D. Not really. They get released, are broken for a bit, then get their numbers gutted and mechanics removed. There is no such thing as "free". Everything a champion does consumes part of their power budget and is reflected in their performance. The issue a lot of these champs present is a high skill ceiling makes their performance drastically different in super high ELO vs normal and low ELO. It makes them impossible to balance them in Masters+ without completely gutting them everywhere else . Look at a champ like Irelia. Still super strong in high ELO while being trash (45% winrate) in Plat+. Ryze, Aatrox, Azir, and others have the same problem
> [{quoted}](name=Eedat,realm=NA,application-id=3ErqAdtq,discussion-id=yZIvfMTw,comment-id=00000000,timestamp=2019-09-25T16:30:15.145+0000) > > Not really. They get released, are broken for a bit, then get their numbers gutted and mechanics removed. There is no such thing as "free". Everything a champion does consumes part of their power budget and is reflected in their performance. The issue a lot of these champs present is a high skill ceiling makes their performance drastically different in super high ELO vs normal and low ELO. It makes them impossible to balance them in Masters+ without completely gutting them everywhere else . Look at a champ like Irelia. Still super strong in high ELO while being trash (45% winrate) in Plat+. Ryze, Aatrox, Azir, and others have the same problem I agree on the gutting numbers, but removing mechanics is by far less frequent and has been done only sporadically. The "power budget" philosophy is perfectly fine in theory, but I haven't seen any realistic application of it. Contrary to what you have written, there are real "free" things that champion gets by virtue of (1) existing; (2) putting a single point in a skill; (3) upgrading an ability (somehow); (4) a combination of the previous. I'm not against that. I'm not against Katarina R getting free Grevious Wounds or Yasuo R getting free Armor Pen. No, that (alone) is not a problem. Katarina Grevious Wounds in R was never even accounted for in power-budget considerations. Again, that fact (alone) is not a problem. But you can't reason in a vacuum. I like Kai'Sa, really. It's one of the few ADCs I enjoy playing and watching. But Kai'Sa, honestly, just has too much in her kit for an ADC. It's overloaded and that's a fact. Compare Kai'Sa to pretty much any "old" ADC and you will immediately see the difference. Take pre-rework Graves (if anybody remembers that such a thing existed and was an actual ADC) and compare it to Kai'Sa. Or even Vayne. Kai'Sa is basically Vayne 2.0. Champion kits being overloaded has little (read: nothing) to do with winrates. Camille has probably the most overloaded kit in all LoL. Is she broken? Right now, no. But who cares? Overload and winrate are uncorrelated, if not straight up independent, variables. You can have the most overloaded kit in history and still have trash winrate across all elos. Because numbers dictate winrate much, much more than kit overload. Especially in the so-called "damage meta". Akali was gutted and had one of her most frustrating mechanics removed. She still is more overloaded than old champions like Annie, Garen, Tristana. Where is the power budget constrain? Akali still has a greater power budget than Annie (and many others, Annie is just an example). And I don't see Riot removing mechanics and buffing numbers in the near future. Don't let me get started on Irelia, because she was omni-present in pro-play for too damn long and deserved all the nerfs she got. It was just absurd seeing Irelia every single game, except when she was banned. Riot messed up big time with her, only Kalista (during her "golden age"), old Kassadin (yeah, the 98% banrate shit, was it S2?), Lulu, S5 Ryze (the "perma-W + spell spam" iteration), and Maokai (pre-rework on R) were, historically, so much disgusting and meta-warping. In S9 "holy mountain" of champions who dominated pro-play, Irelia is right there, alongside (reworked) Aatrox, (reworked) Akali, and pre-nerf Sylas and Kai'Sa (due to almost all other ADCs being subpar). Though, Sylas primarily due to its R being a "meta-ability" (instead of a regular ability), so basically Sylas can be worth 2 champions even when slightly behind, ignoring any power-budget constrain the original champion had with that ability in mind, think Kennen R on Sylas, Kennen (power-budget wise) sacrificed almost everything for that ult (at least pre-nerf), Sylas can happily ignore that and use it. I can only wish that Riot could create a new champion with some mechanics (a "kit loading" comparable to the average of the kit loading we have now, averaging across all champions) and just fine-tune the numbers from there. Instead, they create a champion with an overloaded kit and then need to gut the numbers of that kit not to exceed an imaginary power-budget, which is anyway higher than 20-30% of the champion roster.
: We should need to sacrifice something to build something else
I fully agree with that design principle and I think that it will move the game in a healthier direction. The same principle, in theory, should also be used when designing a champion kit: between (A, B, C), pick two. Unfortunately, newer champions basically get (A,B) for free, can do C with some effort and can also do that extra D. Old champions were a lot more "linear", which doesn't necessarily mean that they lacked any depth. There was a user who compiled a document (and then posted it, [here](https://boards.na.leagueoflegends.com/en/c/gameplay-balance/WxAXtXlR-an-analysis-of-the-notion-of-overloaded-kits-collaborative-google-docs-file)) showing which champion kit was actually overloaded (spoiler: Camille and Ekko are _really_ overloaded). Newer champions, by far, outclassed older ones, except perhaps some reworks which are basically new champions, like Aatrox. Unfortunately, I highly doubt Riot will follow this direction.
: The problem of tank meta wasn't tanks being unkillable, but them doing too much damage. When you can't die and do insane damage then you automatically win. I would prefer a tank meta still tbh. This shit with 3 skills killing you just doesn't require any thinking. Its not engaging to play something this braindead. If the damage was lower, the game would be so much better. Strategy would actually be a part of the game again.
> [{quoted}](name=Shiroyashayami,realm=EUNE,application-id=3ErqAdtq,discussion-id=iHeFMl0J,comment-id=0001000100010001,timestamp=2019-09-25T13:29:46.111+0000) > > The problem of tank meta wasn't tanks being unkillable, but them doing too much damage. When you can't die and do insane damage then you automatically win. I would prefer a tank meta still tbh. This shit with 3 skills killing you just doesn't require any thinking. Its not engaging to play something this braindead. If the damage was lower, the game would be so much better. Strategy would actually be a part of the game again. Exactly. Which is almost paradoxical to think that the biggest issue in the tank meta was (wait for it...) _damage_. What is commonly called "tank meta" was really a "damage+tank meta". So, by removing the ability for tanks to (reasonably) survive, we are left with only "damage" meta. Back then, the common reasoning for having a (tank+damage) meta in the first place was that tank "needed" damage, say for, e.g., wave control, being a threat to the backline, etc. Except for wave control, that reasoning was basically nonsense. Wave control could have been resolved by having two separates damage tables: one against minions, the other against champions. This way, damage against minions could have been fine-tuned separately from damage against champions. Of course, since champions in LoL are/were (quite literally) "coded as minions", this solution, while definitive, would have required major effort. So it was never implemented. And the rest is history.
: Keep telling yourself that. An overloaded kit that makes a champion banned in every second game is a problem and should be looked at.
> [{quoted}](name=Shuriman God,realm=EUW,application-id=3ErqAdtq,discussion-id=FqOs7NRv,comment-id=00140000000000000000000000000000,timestamp=2019-09-25T06:42:23.804+0000) > > Keep telling yourself that. An overloaded kit that makes a champion banned in every second game is a problem and should be looked at. The problem is that, according to Riot own criteria of balance, a 51% banrate champion doesn't raise a flag due to its WinRate not being higher than a certain threshold. I have explained why Riot criteria is critically flawed in my previous post in this topic. And no, contrary to a minority belief, this isn't an issue due to Yasuo being Yasuo: any champion in the same situation (high banrate, good pickrate, WR lower than 52%) would not be considered for balance. This is the real problem. So they won't look at him. Honestly, I will be very surprised if Riot decides to touch Yasuo, even minimally.
: > [{quoted}](name=Ulkusus,realm=EUW,application-id=3ErqAdtq,discussion-id=FqOs7NRv,comment-id=0016,timestamp=2019-09-24T08:25:58.073+0000) > > Apparently 6-7x average ban rate (as given in [this dev post](https://nexus.leagueoflegends.com/en-us/2019/05/dev-champion-balance-framework/) here ) is not enough to change him, he would need a higher win rate for that. > > Whyever a constantly high banrate isnt also a reason to change a champion is beyond me though. **52.5%WR at 5x ABR** Average Ban rate is 7% also. Yasuo meets the ban rate but not the win rate.
Very nice post, Linna Excel, I pretty much agree with you. An important point you made is that also Yasuo mains are negatively affected by it. Moreover, people who would like to main Yasuo ("Yasuo trainee", i.e., neither a first-timer nor a true main) really have a hard time doing it, due to the high ban rate. Which brings us back to Riot criteria: > [{quoted}](name=Antenora,realm=EUW,application-id=3ErqAdtq,discussion-id=FqOs7NRv,comment-id=00160000,timestamp=2019-09-24T09:49:05.449+0000) > > **52.5%WR at 5x ABR** > Average Ban rate is 7% also. > Yasuo meets the ban rate but not the win rate. This criteria (btw, the criteria for "Skilled Play" is basically the same: **52% WR at 5x ABR**, so this applies to almost-all Elos ranges, excluding 0.1% of player-base and pro-play) has a critical flaw. Winrate and banrate are not independent: in fact, banrate seriously affects winrate (to the point of making winrate a useless metric). For increasing banrate, the winrate threshold for "Skilled Play" should decrease accordingly. Thus, extremely high banrate, like the case we have with Yasuo right now, would at least "raise a flag". Riot decided to use a single WR-threshold (at 5x ABR), but in reality this is actually a function: for each value of the banrate, there exist a corresponding WR-hreshold. Also, you don't need to use an integer multiple of ABR, that's just an arbitrary simplification (which leads to stupid results). If ABR=7%, the (5,6,7) multiples are (35, 42, 49), so if a champion has 48.999% banrate, it will still fail to fall into the 49% "bin" and will be judged with a threshold thought for 6xABR, when it is actually extremely close to the 49% bin. What we have now is even worse: a champ with 34.999% banrate, who fails the 5xABR check, will be judged according to the nominal criterion, just like a champion with 10% banrate. Nonsense. There is no need to quantize a function with steps as big as ABR, only a lower cutoff is needed.
: …. that's missing the entire point of EVERYTHING I've said. Unless you are the top of the top, 0.1% of league players, the "impenetrable wall" you mention does not matter. Players make mistakes, even the hardest counterpicks can be won against unless both of you are robotically perfect players. Pro players show games winning against counterpicks ALL the time. The exact thing you said about matchups works the opposite of how you tried to use it, my man. Basically, the entire thing boils down to just getting better. Every single aspect of this game has some way you can work around it, you may not be able to do it 100% solo, but that's why its a team based game. And just because I don't think you should be able to ban every difficult matchup and hand yourself an easy lane for every game, doesn't mean that I shouldn't believe in banning champs period. That's just not good logic. Everyone has their own reasons for banning a pick, but to expect to be handed multiple bans because you cant deal with matchups? That's just a massive crutch, and actually will more than likely make you a worse player in the long run.
> [{quoted}](name=BongMage420,realm=NA,application-id=3ErqAdtq,discussion-id=PfENY8er,comment-id=000000000000000000000002000000000000,timestamp=2019-09-06T09:00:02.358+0000) > > Unless you are the top of the top, 0.1% of league players, the "impenetrable wall" you mention does not matter. Players make mistakes, even the hardest counterpicks can be won against unless both of you are robotically perfect players. Actually, only one of them needs to play "robotically perfect" to survive the matchup, the other one, favored by the matchup, can play a bit less than perfect. The number of bans affects everyone in the game, so you cannot ignore pro-play. In fact, pro-play was the main concern for Riot before the introduction of the current system of 10 bans. It is also why, at pro-play, they have a strategically-deeper ban phase ("snake ban"). > Pro players show games winning against counterpicks ALL the time. Let's talk pro play then. In Season 5 MSI, the Azir vs Zed matchup was horribly one-sided against Azir. Check Febiven Zed vs Faker Azir. And yes, SKT still won the game, doesn't mean the 1v1 matchup wasn't decided by draft (and by highly-skilled piloting of Zed, of course). In the Worlds 2015 patch, Ryze was the best midlaner, period. It was the time when Ryze could do the perma-W prison + spell spam. So much so for "playing robotically perfect" when that version of Ryze simply lacked _**any**_ counter-play, perfect or not. The very definition of "one-sided", which is the worst of what a counter-pick can become. I can go on as much as you want. What you see right now is a side-effect of the current one-shot meta. Of course, when everyone can oneshot everything, counter-picks don't seem that relevant, because it's like everyone has a nuke and so most of the time the outcome is decided by the one who nukes the other first. This tends to create a less skilled environment, mind you, since it is based on capitalizing on _one_ mistake, instead of timely exploitation of multiple mistakes (which requires much more skill). Go back in time a bit and you will see what counter-picks really did. I have already given you plenty examples, but let me add one more: they caused lane-swapping to avoid the matchup entirely. Then, the team had to invade in order to place a ward so to avoid the enemy team predicting the swap and swapping themselves. Counter-picks, for pro-players, were a real thing. > The exact thing you said about matchups works the opposite of how you tried to use it, my man. Basically, the entire thing boils down to just getting better. Every single aspect of this game has some way you can work around it, you may not be able to do it 100% solo, but that's why its a team based game. The psychological aspect of banning is mainly aimed to solo scenarios. It is obvious that there are work-arounds when you introduce more champions into the equation. Also the enemy team has work-arounds to your work-arounds, by the way. > And just because I don't think you should be able to ban every difficult matchup and hand yourself an easy lane for every game, doesn't mean that I shouldn't believe in banning champs period. That's just not good logic. Everyone has their own reasons for banning a pick, but to expect to be handed multiple bans because you cant deal with matchups? That's just a massive crutch, and actually will more than likely make you a worse player in the long run. I said "fully-buy" specifically for what you have written, so we don't disagree here. Like me, you are recognizing the existence of two extremes: 0 bans on one hand, too many bans on the other. You think that 2 bans each player, which is a negligible reduction of the champion pool (even without duplicate bans), is past the "too much" limit. I was only pointing out that, again, for some people 10 bans were too much (and voted against its introduction in the game) and they still decide, right now, not to ban anything. > and yes, we'll call my 6 years of playing, ranging from bronze to plat 1 elo, "limited experience" and a "conformation bias" lol It is. Everyone experience is limited and affected by it, since we only see a small part of the big picture. That's why generalizations of your experience is bad, unless it is supported by solid data.
Show more

Xintium

Level 243 (EUW)
Lifetime Upvotes
Create a Discussion