Normal winrates

So I’ve been wondering about our win/losses pie.
image
Obviously the primary use is to judge other players for their playing habits. If they play stronger players, they’re bad people who never return the gift of playing a stronger player. If they play weaker players, they’re bad people who like to gang up on the weak. And ideally winrate and weaker opponents rate should be 50%. But in practice this never works out like that. Stronger you are harder to find stronger opponents. Weaker you are harder to find weaker opponents.

So I want to explore this, I don’t think this has been addressed before.
I used rank histogram data to create this. Winrates over the ranks.

Table view
N Percentiles: 1 25 50 75 99
1376 >25k 0.00 0.19 0.29 0.38 0.54
190 25k 0.14 0.33 0.40 0.47 0.60
229 24k 0.06 0.35 0.42 0.49 0.62
265 23k 0.05 0.35 0.44 0.51 0.66
292 22k 0.06 0.39 0.46 0.51 0.70
338 21k 0.10 0.39 0.46 0.52 0.65
356 20k 0.20 0.39 0.45 0.52 0.69
480 19k 0.21 0.43 0.48 0.53 0.78
426 18k 0.24 0.41 0.47 0.51 0.70
455 17k 0.28 0.41 0.47 0.51 0.69
524 16k 0.28 0.42 0.47 0.52 0.65
561 15k 0.24 0.42 0.47 0.51 0.64
599 14k 0.31 0.43 0.48 0.52 0.70
618 13k 0.31 0.45 0.49 0.52 0.67
724 12k 0.28 0.45 0.50 0.53 0.67
674 11k 0.29 0.46 0.50 0.54 0.69
721 10k 0.30 0.47 0.51 0.54 0.72
685 9k 0.29 0.47 0.51 0.56 0.70
652 8k 0.29 0.47 0.52 0.56 0.76
578 7k 0.33 0.48 0.53 0.58 0.74
612 6k 0.35 0.49 0.54 0.59 0.77
524 5k 0.29 0.49 0.55 0.61 0.80
509 4k 0.32 0.50 0.56 0.62 0.80
417 3k 0.29 0.50 0.56 0.63 0.84
356 2k 0.34 0.52 0.58 0.66 0.88
290 1k 0.32 0.55 0.62 0.71 0.90
235 1d 0.28 0.58 0.66 0.74 0.94
172 2d 0.40 0.58 0.66 0.76 0.96
105 3d 0.41 0.61 0.71 0.81 0.97
52 4d 0.37 0.63 0.74 0.85 1.00
36 5d 0.35 0.70 0.77 0.87 0.96
24 6d 0.40 0.59 0.75 0.82 0.93
13 7d 0.49 0.74 0.80 0.88 0.95
6 8d 0.49 0.64 0.70 0.90 0.96
2 9d 0.76 0.78 0.79 0.80 0.81
1 9d+ 0.95 0.95 0.95 0.95 0.95

Green line is the median. Yellow region covers half of the players - from 25 percentile to 75 percentile. And red region covers 98% of the players - from 1 to 99 percentile. Players with deviation < 100 (but opponents are of all deviations).

Same but for the rate of weaker opponents.

Table view

|1376|>25k|0.00|0.08|0.19|0.32|0.65|
|190|25k|0.05|0.23|0.35|0.46|0.76|
|229|24k|0.00|0.29|0.38|0.50|0.74|
|265|23k|0.00|0.27|0.40|0.53|0.81|
|292|22k|0.03|0.34|0.43|0.50|0.86|
|338|21k|0.08|0.32|0.42|0.51|0.78|
|356|20k|0.11|0.32|0.42|0.52|0.86|
|480|19k|0.07|0.35|0.45|0.52|0.90|
|426|18k|0.08|0.34|0.43|0.50|0.84|
|455|17k|0.07|0.34|0.42|0.51|0.83|
|524|16k|0.10|0.35|0.44|0.51|0.78|
|561|15k|0.06|0.35|0.44|0.52|0.75|
|599|14k|0.17|0.36|0.45|0.52|0.83|
|618|13k|0.11|0.38|0.46|0.53|0.78|
|724|12k|0.09|0.38|0.48|0.54|0.81|
|674|11k|0.10|0.41|0.48|0.56|0.87|
|721|10k|0.11|0.42|0.50|0.57|0.89|
|685|9k|0.10|0.43|0.51|0.59|0.85|
|652|8k|0.09|0.43|0.52|0.61|0.93|
|578|7k|0.12|0.45|0.54|0.64|0.90|
|612|6k|0.13|0.47|0.56|0.65|0.92|
|524|5k|0.10|0.46|0.57|0.67|0.89|
|509|4k|0.11|0.48|0.58|0.68|0.94|
|417|3k|0.06|0.47|0.57|0.71|0.93|
|356|2k|0.17|0.51|0.64|0.75|0.96|
|290|1k|0.13|0.56|0.69|0.80|0.95|
|235|1d|0.11|0.58|0.71|0.82|0.97|
|172|2d|0.18|0.61|0.73|0.83|1.00|
|105|3d|0.28|0.62|0.75|0.84|0.99|
|52|4d|0.28|0.66|0.79|0.89|1.00|
|36|5d|0.26|0.77|0.82|0.88|0.97|
|24|6d|0.21|0.45|0.74|0.88|0.95|
|13|7d|0.39|0.72|0.81|0.88|0.98|
|6|8d|0.21|0.53|0.67|0.84|0.97|
|2|9d|0.68|0.69|0.71|0.72|0.73|
|1|9d+|0.95|0.95|0.95|0.95|0.95|

I didn’t test it for bugs much (though we did find one bug there, hehe) but looks ok.

4 Likes

Bumping these lovely graphs. I’m surprised at how bloated the red zone is on the lower graph: how do so many players lose nine out of ten games against weaker opponents?!

We are weak players; we lose by weaker players. We exist.