If you have played 1000 games and are still worse than the average player

Competitive Discussion
21

2160 to 849

this is what I dropped my first season, season 7.

What you say may be true. It just seems implausible that I earned 2160 in QP while leveling from 1 to 25 and then in a matter of a month or two, I was 60% worse as an Overwatch player.

22

It’s not always perfect. Many factors could have made the system think you are a better player. Changing metas, hero nerfs and buffs, queuing with a friend who is better than you, thus winning you more games, etc.

The highest % there is gold. I’ve gotten placed lower and way higher than that

Your first games after your first placements, you also get more and lose more sr per win/loss to quickly find where you should be. Its possible you got on a losing streak to start your first season and crashed hard

23

didn’t happen in the middle of season 7, or did it? I don’t think so, but you tell me.

was there a game changing nerf/buff in season 7?

solo queued all the way up to xp lvl 30…at least.

24

Mercy’s change happened in s7 i believe

25

this seems very likely to me, thanks for your time!

1 Like
26

I believe it starts off at around 80 sr and slowly goes down to 20-35 within the span of 15 or so games. I have a file on my pc where i tracked my first games on an alt and i remember the sr was insane before it hit the normal increase/decrease per win/loss

1 Like
27

I somehow got placed in low plat during season 3 (my first OW season) and first fps I had played in YEARS.

Needless to say I lost over 1000 SR. The problem is 2000 SR in QP and 2000 SR in comp are 2 completely different animals. The only people who take QP seriously are just people who play QP and nothing else. REgular comp players generally dont take QP seriously at all.

Your initial placement is based off of 10 comp games with the initial sr based off your QP sr…and as I mentioned before its just not accurate at all.

You can start at 2300ish QP mmr go 7-3 in your placements and get placed in low plat only to face much better competition versus QP.

Most people play QP. It’s going to have a lot more players and thus more food going into the food chain. Average comp players will stomp average QP players generally speaking.

28

Can you cite yours.?

29

already presented my evidence Proteus :stuck_out_tongue:

30

From my understand, your hypothesis is that if your SR doesn’t move much for over 1000 games (say you win-lose-win-lose non-stop for 1000 games), that your MMR’s volatility drops so low that it becomes literally impossible to move out of your SR at that point?

I’m not an expert with rating systems and w/e, but I have a decent understanding of mathematics and I’m fairly certain the system doesn’t work that.

Here are my views…

The relation between SR and MMR sounds similar to a mean-reverting process in the sense that the farther your SR is from your MMR, the faster it will catch up.

Mathematically, this can be done in many ways either via known models (OU process, GARCH, etc.) or ad hoc with discrete adjustments. Considering the level of mathematicians Blizzard employs (I did look it up), these guys are probably not random dudes and have tried multiple models to find the one that worked best. What “works best” can be inferred relatively easily through multiple metrics:

  • % win rate of players gives you a good idea of how fair the system is
  • How often games go into Overtime (more is better)
  • How often games go into Round 2, 3, 4, etc. (more is better)
  • How often games are stomps (less is better)

Let’s just assume Blizzard did their homework and look at SR / MMR. In a mean-reverting process, your process tries to come back to the mean. This can be slow or fast depending on your parameters. Here’s an example of how such a function would work (not what Blizzard is using, just an example):

SR(t) = SR(t-1) ± 25 + a * [MMR(t) - SR(t)]
where a is the mean reverting parameter

  • The bigger the difference between MMR and SR, the bigger the adjustment
  • When your SR = MMR, you win or lose your usual 25 SR per game
  • When SR < MMR, you win/lose your usual 25 SR + some bonus (catch-up)
  • When SR > MMR, you win/lose your usual 25 SR - some penalty (catch-up)

At first glance, it seems like your SR is stuck at your MMR. Generally speaking, yes! And it’s a good thing! It means you’re playing fair games often and your SR doesn’t need adjustments. But you might be wondering “What if I improve? Will my MMR keep pulling on my SR to bring it back down?”

The way MMR varies sounds similar to a stochastic volatility model in the sense that your MMR’s volatility will expand and retract based on recent games.

The cool thing about 2 speed system (SR and MMR) is that they can move at different speeds. MMR is more of a local rating while SR is a global one. MMR closely reflects the way you played recently. When you win (or lose) a single game, your MMR varies a little bit faster than your SR (say ± 30 instead of 25). But if you win (or lose) a few games in a row, then your MMR progressively increases (or decreases) faster (say ± 30, 40, 50, 75, 100, etc.) This is due to MMR’s volatility inscreasing (or decreasing) as it registers you are on a win (or loss) streak.

After winning say 9/10 games tonight, your SR might only go up 240 SR but your MMR might have gone up 500 MMR. And every following game, your SR will try to catch up to your MMR and thus you will win 30-35 SR instead of the usual 25. Boosters have reported winning upwards of 60 SR after a couple games won in a row, suggesting that MMR adjusts rather rapidly (less than 20 games) and can greatly impact your SR.

The other good thing about this two speed system is that if you have a lucky night and win a lot of games, the system will give you the benefit of the doubt by raising your MMR a lot so that you can “cash in” by winning at least 50% of your games at your new high MMR. And if you were simply lucky, then although you benefited from your high MMR for some time, you’re not going to fall 1000 SR the day after.

So if you play at 2000 SR on day one and 3000 SR on day two, the system may be able to give you appropriate games rapidly for each day, but your SR ± 200 SR instead of ±1000 MMR.

You can make a work analogy:
If you work hard for a few weeks (games), your boss will notice and give you more responsibilities (higher MMR) and some small bonuses every week (SR catching up). If you can’t keep up, then he stops sending you more responsibilities (lower MMR) and the small bonuses stop coming in. But if, when he gave you more responsibilities, you easily pulled through and kept on asking for more, he recognizes your ability and gives you a full on promotion (higher SR).

1 Like
31

when you play qp it is an extremely unstructured environment. making any kind of prediction (rank) very unpredictable. so when in qp you were getting average stats against these players. they were probably meaningless stats. (damage that doesnt secure a kill, hitting everyone on their team once for more elims, etc.) but once you go into comp and play against average people. you fall extremely quickly cause these people somewhat know what they are doing. while you are still clueless about most things. that damage you were doing doesnt matter because they just ran away and got healed. you hitting all those people didnt matter because they all survived without being focused. so the game realizes its mistake and starts putting you lower and lower untill you hit your real rank. where everyone is doing what your doing.

32

This is not what I’m saying. What I’m saying is that, assuming your progress as a better Overwatch player is a slow and steady process, and it is also impossible to get worse at Overwatch. What I witnessed my SR do seems to be bizarre.

33

How is it impossible to get worse at Overwatch? People get worse at everything all the time. And worse is also relative in a competitive environment: If you don’t improve, everyone else around you does and you end up a worse player relative to the competition.

34

so are we all perpetually improving or not? you kinda contradicted yourself there

35

…

On average, players improve. That doesn’t mean you do as an individual.

36

ok well I’m telling you I think my play only got better as time went on. perhaps there is something I didn’t notice about my play, and the way the SR system is designed had a 0% probability of incorrectly assigning me a 2150SR after my first 10 placement matches

37
  1. Play more than 10 games
  2. Just because you think you play well/better doesn’t mean you do.
38

If you need more than 10 games to get an accurate SR, then Blizzard assigns inaccurate SR, which is the point of my entire post.

39

If you can’t see the curvature of the Earth, then it isn’t a sphere.
I’m done, you’re either a troll or just delusional.

40

so that’s what a boxer says after he accidently knocks himself out lol