High win % SR penalty (with Data!)

Competitive Discussion
Several blue posts have indicated that SR chases MMR. This implies that if SR is above MMR, less SR than MMR will be gained on a win, and more SR than MMR will be lost on a loss. This is an SR debuff. If SR is below MMR, more SR than MMR will be gained on a win, and less SR than MMR will be gained on a loss. This is an SR buff. This leads to an asymmetry: if a player wins many games, his SR will be above his MMR and he will gain less SR on a win than he loses on a loss. This will persist until his win percentage falls to approximately 50%. (The reverse happens if a player has a very low win %).

To summarize: MMR acts as an anchor/flotation device for SR.

Disclaimers

Neither Blizzard nor I approve of sharing accounts.

I do not approve of Bronze to GM series. They give interesting data, but this is not worth the hundreds of games that they ruin.

Please do not speak of specific banning actions in this thread. That will cause the thread to be deleted, which would make me sad.

The Data

https://docs.google.com/spreadsheets/d/1FLEzkX4AMFsO2tU2kNOxyqO4SCJ2kDYI8qa0QGk4V9A/edit?usp=sharing

This data comes from a Stevo Bronze to GM series, in which he played only Symmetra. The account had a well-establish bronze MMR when Stevo started playing on it. The green dots show SR gained on win, and the red dots show SR lost on loss. The black traces show a 31 game lagging average of SR changes. The blue trace shows a 31 game lagging average of win percentage. In the beginning there is a noticeable SR buff, which has been removed from the game (the data is from season 5). The middle section shows approximate SR gained on win of 22, and SR lost on loss of 28. This is where the SR/MMR discrepancy and SR debuff is in full force. Toward the end, there are two times where the SR debuff goes away. Both of these times, Stevo’s win percentage had recently dropped to 50%, which allowed his MMR to catch up to his SR.

How to remove the high win % penalty

Abolish SR. Make MMR public. Deal with problems like rank camping some other way than SR decay.

Alternatively, it would be possible to do a cludgy fix in which a small SR buff is added to counter act the natural debuff from the algorithms in specific cases. I'm not found of this solution, but I expect it to be the one that Blizzard takes.

Isn't this just a performance effect? (Edit based on comments)

In general, it is difficult to distinguish whether an SR gain asymmetry comes from the MMR/SR algorithms or performance modifiers without having more data than non-Blizzard employees can reasonably gather. So reasonable people can disagree. However, I don't think that performance was the issue here, for the following reasons:

1) Stevo's win percentage from when he first hit Diamond till when he hit Master was 75.9%, which implies that his performance was off the charts. However, he had a substantial SR debuff at this point.

2) The SR debuff went away at high ranks, when Stevo would presumably have lower performance metrics due to stiffer competition. This is backwards from what we'd expect if performance metrics were the reason for his SR debuff.

3) Toward the end, the SR debuff went away shortly after Stevo's win percentage dropped to 50%. It did this twice. I expect that Stevo's performance is extremely consistent, and he just got unluckly (and the competition got stiffer) for a while. This would imply that what happened is that MMR caught up to SR, and not that his performance took a dive.

Regardless, if it is indeed caused by performance rather than MMR/SR, it is still a bug which requires Blizzard's attention.

P.S.

If you have data you would like me to run through my analysis, please post it to google docs in a similar format and share it with me.

If you have data which includes all performance metrics (from the score screens at game end) for many games played, share that with me as well.
Part of the problem is the "streak" modifier is gone now too. They removed this to alleviate "boosting" and "throwing" to high/low MMR.

However, now when a good player creates a smurf/alt it takes 2-3x longer to raise it to the MMR it belongs. Also the same can be said in the opposite direction. Going 0-5 no longer increases how fast you're losing MMR at high ranks (I'm thankful for that at times because I have really bad days) but it means you get crushed for hours and hours and SLOWLY decline. I went from 4048 to 3951 on an 0-4 streak and now am climbing back up. Honestly I'm surprised I didn't lose more SR because I was playing TERRIBLE this weekend.
So, even a GM playing below his real MMR have +20 per win and -30 from losses. GJ Devs, great system
There are many good reasons to have a parallel rating system like SR which exists on top of MMR. As long as they never deviate too far from one another, you can pair the visible one with rewards or decay, and in general that's mostly fine. It gives you a few levers and knobs to tweak since it doesn't impact your One Source Of Skill Truth, MMR.

There are similarly good reasons why you wouldn't want to show MMR. For example, in some cases, MMR can go down with a win. It's rare, but it can happen especially at the top end. If you play a really good game, your MMR may not change at all. What that really means is that you are performing at a level consistent with your estimated ability, but players' expectations are different and they believe "I did well = MMR goes up." It basically becomes impossible to message. Also, showing MMR is somewhat risky for developers because if they ever have to change the math model (happens all the time especially early in a game's lifecycle), that can introduce thrash or feel jarring, whereas a parallel rating system can soften that effect or perception.

09/18/2017 10:47 AMPosted by Murdinn
So, even a GM playing below his real MMR have +20 per win and -30 from losses. GJ Devs, great system


I believe that's the wrong way to read this Bronze-to-GM example. MMR moves more slowly than SR by design. According to the system, this is a Firmly Bronze player. Uncertainty is very low that he is not Bronze because he's been Bronze for a very long time. The system doesn't know that there's a different player behind the controls, it only knows that he's suddenly winning more. Therefore, it needs to gradually branch out as his MMR climbs. In TrueSkill-style systems like Overwatch's, uncertainty narrows fairly quickly and doesn't expand again with the same speed.

If you were playing with a fresh account, it would be able to send you up to GM-land pretty fast. But because there was an established player in this case, it's going to be more hesitant to catapult someone up the ladder since prior data suggests this player was once finding competitive games at the bottom and now that appears to no longer be the case. It needs to be sure the more recent games aren't outliers or noise.
09/18/2017 01:41 PMPosted by ExcaliburZ
There are many good reasons to have a parallel rating system like SR which exists on top of MMR. As long as they never deviate too far from one another, you can pair the visible one with rewards or decay, and in general that's mostly fine. It gives you a few levers and knobs to tweak since it doesn't impact your One Source Of Skill Truth, MMR.

There are similarly good reasons why you wouldn't want to show MMR. For example, in some cases, MMR can go down with a win. It's rare, but it can happen especially at the top end. If you play a really good game, your MMR may not change at all. What that really means is that you are performing at a level consistent with your estimated ability, but players' expectations are different and they believe "I did well = MMR goes up." It basically becomes impossible to message. Also, showing MMR is somewhat risky for developers because if they ever have to change the math model (happens all the time especially early in a game's lifecycle), that can introduce thrash or feel jarring, whereas a parallel rating system can soften that effect or perception.

09/18/2017 10:47 AMPosted by Murdinn
So, even a GM playing below his real MMR have +20 per win and -30 from losses. GJ Devs, great system


I believe that's the wrong way to read this Bronze-to-GM example. MMR moves more slowly than SR by design. According to the system, this is a Firmly Bronze player. Uncertainty is very low that he is not Bronze because he's been Bronze for a very long time. The system doesn't know that there's a different player behind the controls, it only knows that he's suddenly winning more. Therefore, it needs to gradually branch out as his MMR climbs. In TrueSkill-style systems like Overwatch's, uncertainty narrows fairly quickly and doesn't expand again with the same speed.

If you were playing with a fresh account, it would be able to send you up to GM-land pretty fast. But because there was an established player in this case, it's going to be more hesitant to catapult someone up the ladder since prior data suggests this player was once finding competitive games at the bottom and now that appears to no longer be the case. It needs to be sure the more recent games aren't outliers or noise.


This is the one issue with TS especially when you begin adding extensions in. It's unable to interpret a quick rise in skill. It takes HUNDREDS of games of you performing well (and winning) to increase your MMR dramatically. Of course, this isn't USUALLY the case anyways.

It's very very accurate for your average person, but when someone much better than you takes over your account of course it's going to have the typical issues that we see in the "bronze to GM" quick pushes.

I think there's some tweaking that can be done but overall taking a bronze account to GM SHOULD take several hundred games because that's how the average player would actually be increasing.
09/18/2017 01:53 PMPosted by Exziled
There are many good reasons to have a parallel rating system like SR which exists on top of MMR. As long as they never deviate too far from one another, you can pair the visible one with rewards or decay, and in general that's mostly fine. It gives you a few levers and knobs to tweak since it doesn't impact your One Source Of Skill Truth, MMR.

There are similarly good reasons why you wouldn't want to show MMR. For example, in some cases, MMR can go down with a win. It's rare, but it can happen especially at the top end. If you play a really good game, your MMR may not change at all. What that really means is that you are performing at a level consistent with your estimated ability, but players' expectations are different and they believe "I did well = MMR goes up." It basically becomes impossible to message. Also, showing MMR is somewhat risky for developers because if they ever have to change the math model (happens all the time especially early in a game's lifecycle), that can introduce thrash or feel jarring, whereas a parallel rating system can soften that effect or perception.

<span class="truncated">...</span>

I believe that's the wrong way to read this Bronze-to-GM example. MMR moves more slowly than SR by design. According to the system, this is a Firmly Bronze player. Uncertainty is very low that he is not Bronze because he's been Bronze for a very long time. The system doesn't know that there's a different player behind the controls, it only knows that he's suddenly winning more. Therefore, it needs to gradually branch out as his MMR climbs. In TrueSkill-style systems like Overwatch's, uncertainty narrows fairly quickly and doesn't expand again with the same speed.

If you were playing with a fresh account, it would be able to send you up to GM-land pretty fast. But because there was an established player in this case, it's going to be more hesitant to catapult someone up the ladder since prior data suggests this player was once finding competitive games at the bottom and now that appears to no longer be the case. It needs to be sure the more recent games aren't outliers or noise.


This is the one issue with TS especially when you begin adding extensions in. It's unable to interpret a quick rise in skill. It takes HUNDREDS of games of you performing well (and winning) to increase your MMR dramatically. Of course, this isn't USUALLY the case anyways.

It's very very accurate for your average person, but when someone much better than you takes over your account of course it's going to have the typical issues that we see in the "bronze to GM" quick pushes.

I think there's some tweaking that can be done but overall taking a bronze account to GM SHOULD take several hundred games because that's how the average player would actually be increasing.


You hit the nail on the head. TS aims to hone in on a player's window and it actually does it relatively quickly and fairly accurately for the vast majority of players.

A person does not just wake up one morning and become a vastly improved player at anything. A rise in skill takes time and TS is able to track it pretty well for the 99% case. As you said a different player taking the controls is a huge 0.0001% outlier and is hard to deal with for any system as it essentially negates the heuristics being used on the MMR windowing.

Also a lot of times there is a large difference between how people "feel" they are playing and the raw data metrics being collected on how a person actually played. People typically remember the outlier events a lot more than the everyday norm but the data collection system remembers all equally and applies them as such.

It can definitely feel like a grind but that is totally normal given that your skill level rise is gradual with several plateaus, peaks, and valleys. Hence your SR and MMR mirror those same ups and downs and MMR is the trailer so it has a bit of smoothing on the quick changes. Its also important to keep in mind as said above MMR can actually not change or go slightly down even on a win (not typical to go down though)
Oh look, another data collection attempt that attempts to make conclusions about a performance-based SR system without actually collecting performance stats such as elims, deaths, damage, and etc. What a great argument!

Your analysis is nothing more than looking at the surface. Not only does it not include other factors, but you don't even pretend that those factors exist.

This is the problem with people like you who don't understand statistics. You have 0 understanding of statistical thresholds. Thefact that there are data between +30 and +16 and -34 and -20 implies that performance is a factor, which is something you're not factoring in. Please get a clue before you erroneously post about something you have no understanding of.
09/18/2017 02:08 PMPosted by mrjman

You hit the nail on the head. TS aims to hone in on a player's window and it actually does it relatively quickly and fairly accurately for the vast majority of players.


Just out of curiosity: Why makes you believe that's true? I haven't seen error rates posted anywhere for SR calculations.
09/18/2017 02:21 PMPosted by MauriCleric
Oh look, another data collection attempt that attempts to make conclusions about a performance-based SR system without actually collecting performance stats such as elims, deaths, damage, and etc. What a great argument!

Your analysis is nothing more than looking at the surface. Not only does it not include other factors, but you don't even pretend that those factors exist.

This is the problem with people like you who don't understand statistics. You have 0 understanding of statistical thresholds. Thefact that there are data between +30 and +16 and -34 and -20 implies that performance is a factor, which is something you're not factoring in. Please get a clue before you erroneously post about something you have no understanding of.


Thanks for clearing it up for everyone and showing how to understand it all.
Oh wait, all you did was insult someone. GJ.
09/18/2017 02:44 PMPosted by JOHNNYMATRIX
09/18/2017 02:21 PMPosted by MauriCleric
Oh look, another data collection attempt that attempts to make conclusions about a performance-based SR system without actually collecting performance stats such as elims, deaths, damage, and etc. What a great argument!

Your analysis is nothing more than looking at the surface. Not only does it not include other factors, but you don't even pretend that those factors exist.

This is the problem with people like you who don't understand statistics. You have 0 understanding of statistical thresholds. Thefact that there are data between +30 and +16 and -34 and -20 implies that performance is a factor, which is something you're not factoring in. Please get a clue before you erroneously post about something you have no understanding of.


Thanks for clearing it up for everyone and showing how to understand it all.
Oh wait, all you did was insult someone. GJ.


Yeah because I totally didn't explain that including performance stats would actually make the argument valid.

It's almost as if you didn't read what I said at all.
oh i read it. it didnt make what you said any more valid than the other p. care to try again? This time without insulting people..cuz...its hard to hear through the bully leaking out all out ya neck...
09/18/2017 02:43 PMPosted by Mindaika
09/18/2017 02:08 PMPosted by mrjman

You hit the nail on the head. TS aims to hone in on a player's window and it actually does it relatively quickly and fairly accurately for the vast majority of players.


Just out of curiosity: Why makes you believe that's true? I haven't seen error rates posted anywhere for SR calculations.


Well like any system that attempts to rank players there are always tradeoffs. I don't 100% know that OW even uses TS, but there have been studies from other games with team play that show TS converging very quickly and showing the accuracy of the perceived outcome vs true outcome to be very good (and even more so with strong convergence). My guess is that OW is using some form of modified TS that has modifiers for personal performance among other things to try and hone it in a bit more.

To my knowledge no information has been released about what OW specifically uses or the accuracy. I was more talking about TS in general rather than OW.
09/18/2017 02:21 PMPosted by MauriCleric
Oh look, another data collection attempt that attempts to make conclusions about a performance-based SR system without actually collecting performance stats such as elims, deaths, damage, and etc. What a great argument!

Your analysis is nothing more than looking at the surface. Not only does it not include other factors, but you don't even pretend that those factors exist.

This is the problem with people like you who don't understand statistics. You have 0 understanding of statistical thresholds. Thefact that there are data between +30 and +16 and -34 and -20 implies that performance is a factor, which is something you're not factoring in. Please get a clue before you erroneously post about something you have no understanding of.


Shoe, is that you? Anyways, I expect that Stevo's performance is extremely stable and extremely high for when he was in diamond and below. It is theoretically possible that the issue is not with SR/MMR but with a bug in the performance calculation. For example, Stevo likes to use shield generators in the lower ranks. Maybe the performance calculator does not value shield generators highly enough.

However, I do not think that this is the case, because the SR debuff tracked so well with win rate. Whenever his win rate fell to 50%, the SR debuff went away.

I would have tracked Stevo's performance numbers if I could have, but he frequently exited the game before the stats screen showed. Also, I think performance metrics are more reasonable when players are playing near their proper rank.

Moving on though, I do intend to take and analyze a data set with full performance metrics. I have a new baby though, so I can't play much competitive, and it may take a while. Analyzing data I can do, though, since I can always take a break when she needs to be soothed back to sleep. I would appreciate it if you or anyone else would generate the appropriate data for me.

What I need is this: The complete results from the final score screen for a minimum of 50 games with one hero (the more games the better). This includes elims, deaths, assits, ult numbers, all hero specific numbers, etc. Medal numbers would be nice, but is probably not important. Match time is required. Ideally, the stats should be from a person playing at their proper rank (no bronze to GM series).
09/18/2017 02:21 PMPosted by MauriCleric
Oh look, another data collection attempt that attempts to make conclusions about a performance-based SR system without actually collecting performance stats such as elims, deaths, damage, and etc. What a great argument!

Your analysis is nothing more than looking at the surface. Not only does it not include other factors, but you don't even pretend that those factors exist.

This is the problem with people like you who don't understand statistics. You have 0 understanding of statistical thresholds. Thefact that there are data between +30 and +16 and -34 and -20 implies that performance is a factor, which is something you're not factoring in. Please get a clue before you erroneously post about something you have no understanding of.


I did that for 30ish games back at the end of Season 5 and the stats didn't really correlate to anything specific. I could do super poorly and gain 23 or lose 16. Unless it's a lack or quality of said data - I dunno what to tell ya.

Note: This observation leads me to the opinion of the SR divide being more important than performance. ¯\_(ツ)_/¯

Several blue posts have indicated that SR chases MMR. This implies that if SR is above MMR, less SR than MMR will be gained on a win, and more SR than MMR will be lost on a loss. This is an SR debuff. If SR is below MMR, more SR than MMR will be gained on a win, and less SR than MMR will be gained on a loss. This is an SR buff. This leads to an asymmetry: if a player wins many games, his SR will be above their MMR and he will gain less SR on a win than he loses on a loss. This will persist until his win percentage falls to approximately 50%. (The reverse happens if a player has a very low win %).

To summarize: MMR acts as an anchor/flotation device for SR.


I'm VERY skeptical of the bolded opinion. I have noticed the SR anchor without being anywhere close to 50% winrate (~40%).

Everything else looks to be in line with what I've been saying/found since the end of S5.
09/18/2017 10:44 AMPosted by Exziled
Part of the problem is the "streak" modifier is gone now too. They removed this to alleviate "boosting" and "throwing" to high/low MMR.

However, now when a good player creates a smurf/alt it takes 2-3x longer to raise it to the MMR it belongs. Also the same can be said in the opposite direction. Going 0-5 no longer increases how fast you're losing MMR at high ranks (I'm thankful for that at times because I have really bad days) but it means you get crushed for hours and hours and SLOWLY decline. I went from 4048 to 3951 on an 0-4 streak and now am climbing back up. Honestly I'm surprised I didn't lose more SR because I was playing TERRIBLE this weekend.


I definitely agree that streak bonuses should be removed. They had a tendency to throw people far from their proper rank, through luck alone. However, there should not be the reverse either (a streak penalty). A win should be a win, no more and no less (weighted only by expected win probability).

I actually like the side effect that it took Stevo forever to level up. Even though he ruined more games, it took forever which would make paid services more expensive, and bronze to GM streams more boring to watch. That is, the more costly boosting is, the less of it there will be.
09/18/2017 02:21 PMPosted by MauriCleric
Oh look, another data collection attempt that attempts to make conclusions about a performance-based SR system without actually collecting performance stats such as elims, deaths, damage, and etc. What a great argument!

Your analysis is nothing more than looking at the surface. Not only does it not include other factors, but you don't even pretend that those factors exist.

This is the problem with people like you who don't understand statistics. You have 0 understanding of statistical thresholds. Thefact that there are data between +30 and +16 and -34 and -20 implies that performance is a factor, which is something you're not factoring in. Please get a clue before you erroneously post about something you have no understanding of.


I did that for 30ish games back at the end of Season 5 and the stats didn't really correlate to anything specific. I could do super poorly and gain 23 or lose 16. Unless it's a lack or quality of said data - I dunno what to tell ya.

Note: This observation leads me to the opinion of the SR divide being more important than performance. ¯\_(ツ)_/¯

Several blue posts have indicated that SR chases MMR. This implies that if SR is above MMR, less SR than MMR will be gained on a win, and more SR than MMR will be lost on a loss. This is an SR debuff. If SR is below MMR, more SR than MMR will be gained on a win, and less SR than MMR will be gained on a loss. This is an SR buff. This leads to an asymmetry: if a player wins many games, his SR will be above their MMR and he will gain less SR on a win than he loses on a loss. This will persist until his win percentage falls to approximately 50%. (The reverse happens if a player has a very low win %).

To summarize: MMR acts as an anchor/flotation device for SR.


I'm VERY skeptical of the bolded opinion. I have noticed the SR anchor without being anywhere close to 50% winrate (~40%).

Everything else looks to be in line with what I've been saying/found since the end of S5.


All data like this has the problem of being limited. You used to be low bronze (gratz on the improvement). At the ends of the scale, there tends to be non-50% effects because the of lack of people that are worse (better) at the bottom (top) of the scale. My analysis would not see these effects.
09/18/2017 02:43 PMPosted by Mindaika
09/18/2017 02:08 PMPosted by mrjman

You hit the nail on the head. TS aims to hone in on a player's window and it actually does it relatively quickly and fairly accurately for the vast majority of players.


Just out of curiosity: Why makes you believe that's true? I haven't seen error rates posted anywhere for SR calculations.


08/20/2017 08:36 AMPosted by Kaawumba
How accurate is SR?

On the forums there are generally two factions with respect to SR, those who think that SR is essentially luck and a meaningless value, and those who believe that SR is essentially correct. Neither of these factions have the correct approach. The correct question is to ask how accurate is SR? Clearly top 500 players are better than bottom 500 (see any bronze to GM series or watch low bronze play vs top play), so SR is not completely random, but how accurate is it really? For this discussion, I am assuming that we are talking about a player that plays to win every game, doesn’t share his account, and has played at least 100 competitive games.

There are number of ways to approach this question. One is to start a completely new account, and then play 100 games on the new account and see how it performs compared to the old one. This shows that SR can vary by 1000 SR in extreme cases, and 500 in normal cases (27). There is some evidence that reroll experiments show less variance at higher ranks (28) likely because there are less random variables, such as smurfs, throwers, and inconsistent play.

Another experiment we can do is look at how SR changes between seasons. Even though SR is generally about where it was from the previous season, there is roughly a 900 SR range (in extreme cases) that cannot be explained by previous season ending and placements record (see previous section). Regardless of where this comes from, this implies that SR is not particularly accurate.

Next, any player can see how his SR changes during a season. A range of 500 is completely normal here.

In addition, if an account needs a 55% win rate to maintain SR (11, 12), and if win % changes slowly with rank, then it is expected that this will be an additional source of significant error in SR.

If a player gets noticeably better, it can take a long time to get to the correct rank. If a player maintains a 55% win rate, he will only go up approximately 220 SR/MMR every 100 games. Because there are twelve players in a match, one player's contribution (and ability to carry the match) is limited unless he is playing at a vastly different skill level, so a 55% win rate for a player that is moderately under placed is to be expected.

Put this all together, and we can state that an active and motivated player’s SR is only accurate to +/- 250 SR in normal circumstances and +/- 500 SR in extreme cases. Of course, new players or players that actively break the system can be off by much more.
<span class="truncated">...</span>

I did that for 30ish games back at the end of Season 5 and the stats didn't really correlate to anything specific. I could do super poorly and gain 23 or lose 16. Unless it's a lack or quality of said data - I dunno what to tell ya.

Note: This observation leads me to the opinion of the SR divide being more important than performance. ¯\_(ツ)_/¯

<span class="truncated">...</span>

I'm VERY skeptical of the bolded opinion. I have noticed the SR anchor without being anywhere close to 50% winrate (~40%).

Everything else looks to be in line with what I've been saying/found since the end of S5.


All data like this has the problem of being limited. You used to be low bronze (gratz on the improvement). At the ends of the scale, there tends to be non-50% effects because the of lack of people that are worse (better) at the bottom (top) of the scale. My analysis would not see these effects.


Instead of being described as an anchor, float, or weight perhaps it would be better represented as a rubber band being moved by two fingers. As one finger pulls away it will stretch but create tension. I'm not certain if the rubber has to snap or if at some point the first finger gets reset. In any case, it feels really bad.

I believe that the winstreak modifier was, actually, hiding this from us.

-----------------------SPECULATION-----------------------
Does MMR have some sort of ranking to which it defaults to based on your starting MMR? Where and when does it reset? It pulls from a large data set very quickly for performance adjustments - does it adjust based on your PREVIOUS games MMR value?

So many questions! haha.
I don't think you guys understand the problem with one tricking. He's playing only Symmetra, and at no point swaps even if the team is getting destroyed. He carried hard until gold(thats a given), but when he reached platinum he wasn't crushing players like he was in bronze. After platinum his SR gain was around low 20s you know why? Because plat to diamond players know how to play against Symmetra, they won't blindlessly run into turrets over and over (sometimes people still do in high SR).

It's the loss that negatively impact him more, and for good reason. This game is about strategic adaptation, which is something he isn't doing.
You have to think about the negative more so than the positive because he's going to do worse if his team doesn't do well. You have to think about the math of contribution, it's not like he's playing dps who can completely change the game.
09/18/2017 05:23 PMPosted by Maestro
I don't think you guys understand the problem with one tricking. He's playing only Symmetra, and at no point swaps even if the team is getting destroyed. He carried hard until gold(thats a given), but when he reached platinum he wasn't crushing players like he was in bronze. After platinum his SR gain was around low 20s you know why? Because plat to diamond players know how to play against Symmetra, they won't blindlessly run into turrets over and over (sometimes people still do in high SR).

It's the loss that negatively impact him more, and for good reason. This game is about strategic adaptation, which is something he isn't doing.
You have to think about the negative more so than the positive because he's going to do worse if his team doesn't do well. You have to think about the math of contribution, it's not like he's playing dps who can completely change the game.


In the early games he was still under the influence of the (since removed) early season SR buff. So you can't really interpret those high gains as being performance related. From the time Stevo reached Diamond for the first time, till he reached Master, his win rate was 75.9%. <sarcasm>If only I could play so badly.</sarcasm>

Join the Conversation

Return to Forum