How Getting Good Affects the Game Outcome?

The games after ranking update, no handicap, less than 100 deviation, should be not bots, games with opponents of the same rank. Technically would be nice to filter by time controls but it’s too hard.

For the percent of games finished with certain points we count out of total games including resignations.

19x19

Rank Total Resign Count Timeout Disconnect Median Mean 95 Percentile 0.5 point 1.5 point <=7.5 points <=10 points <=20 points
25k 279 59.50% 24.37% 6.81% 9.32% 49.0 73.9 179.4 0.00% 0.00% 0.36% 0.36% 3.23%
21k 100 48.00% 32.00% 13.00% 7.00% 51.0 64.2 159.3 0.00% 0.00% 3.00% 3.00% 7.00%
20k 149 54.36% 32.21% 8.05% 5.37% 38.0 48.9 107.5 0.00% 0.00% 3.36% 4.03% 9.40%
19k 161 50.93% 39.13% 3.73% 6.21% 43.5 48.8 131.0 1.86% 0.00% 4.97% 4.97% 11.80%
18k 201 45.27% 44.28% 5.47% 4.98% 45.5 53.3 136.1 0.00% 1.49% 2.99% 4.48% 10.95%
17k 267 55.43% 32.21% 5.99% 6.37% 38.0 50.2 130.8 0.37% 0.37% 4.49% 5.99% 11.61%
16k 472 45.55% 45.55% 5.08% 3.60% 36.5 48.8 133.5 0.64% 0.42% 5.08% 6.78% 12.50%
15k 797 47.80% 45.29% 3.64% 3.01% 30.5 40.6 107.5 1.00% 0.75% 7.15% 7.78% 15.06%
14k 1459 47.70% 44.28% 4.87% 2.88% 29.5 38.7 102.5 0.75% 0.89% 6.99% 9.32% 16.24%
13k 2700 48.11% 44.44% 4.33% 3.11% 26.5 34.5 93.5 0.74% 1.11% 7.96% 9.81% 17.41%
12k 3707 48.15% 44.51% 4.96% 2.32% 24.5 34.2 91.5 1.08% 0.73% 8.23% 10.01% 18.24%
11k 5005 48.95% 43.36% 5.53% 2.12% 24.5 32.6 89.0 0.98% 1.08% 8.05% 9.87% 18.04%
10k 7433 53.32% 39.32% 5.27% 2.07% 22.5 30.8 87.4 0.97% 0.89% 7.67% 9.65% 17.92%
9k 9393 53.36% 39.15% 5.83% 1.64% 20.5 28.3 79.5 1.17% 1.03% 8.76% 10.81% 19.28%
8k 10478 56.16% 36.20% 6.13% 1.51% 19.5 26.6 71.5 1.04% 1.10% 8.37% 10.34% 18.10%
7k 14687 58.45% 33.84% 6.12% 1.57% 18.5 25.2 69.5 1.04% 1.01% 8.14% 9.99% 17.80%
6k 16735 61.15% 30.48% 7.03% 1.32% 18.5 24.0 67.5 0.94% 1.01% 7.65% 9.38% 16.44%
5k 18483 60.30% 29.79% 8.32% 1.58% 16.5 22.1 60.5 1.13% 1.08% 7.92% 9.82% 16.90%
4k 21948 60.73% 29.17% 8.51% 1.58% 15.5 20.9 55.5 0.98% 1.21% 8.04% 9.93% 17.54%
3k 19942 61.91% 27.29% 9.34% 1.44% 14.5 19.3 51.5 0.97% 0.96% 8.01% 9.88% 17.43%
2k 19600 63.62% 24.77% 10.45% 1.14% 14.5 18.8 51.5 0.97% 0.96% 7.64% 9.29% 15.97%
1k 14633 61.48% 24.79% 12.90% 0.83% 12.5 17.3 47.5 1.02% 1.13% 8.25% 10.02% 16.89%
1d 10650 62.84% 22.54% 13.82% 0.78% 12.5 16.2 43.5 0.96% 1.05% 7.69% 9.31% 15.93%
2d 3425 61.93% 21.26% 16.32% 0.50% 10.5 14.2 38.5 0.96% 1.20% 8.29% 10.07% 16.03%
3d 1205 70.54% 16.43% 12.53% 0.50% 11.5 15.0 40.5 0.83% 0.83% 5.89% 7.22% 12.03%
4d 467 68.52% 16.70% 14.35% 0.21% 9.5 14.9 38.6 1.07% 1.07% 6.21% 8.57% 12.85%

9x9

Rank Total Resign Count Timeout Disconnect Median Mean 95 Percentile 0.5 point 1.5 point <=7.5 points <=10 points <=20 points
25k 1976 42.00% 48.33% 6.88% 2.78% 15.5 20.6 73.8 1.47% 2.18% 14.32% 17.61% 30.47%
24k 289 45.33% 48.10% 5.54% 1.04% 14.5 20.6 73.5 1.38% 1.38% 12.80% 15.92% 31.14%
23k 414 44.93% 45.89% 7.49% 1.45% 14.0 17.4 45.0 2.17% 2.17% 16.91% 20.05% 29.47%
22k 712 43.12% 49.44% 5.48% 1.97% 13.5 17.8 47.4 2.11% 2.39% 15.45% 19.52% 33.43%
21k 914 40.37% 51.64% 5.47% 2.41% 13.5 18.4 54.5 1.97% 1.20% 15.43% 20.02% 34.46%
20k 1133 43.25% 51.81% 3.35% 1.59% 12.5 16.9 43.9 2.29% 2.82% 16.33% 20.04% 36.10%
19k 1485 43.91% 49.63% 4.58% 1.89% 12.5 16.8 48.5 2.69% 2.22% 17.17% 20.67% 35.89%
18k 1604 42.21% 52.62% 4.05% 1.00% 11.5 15.8 44.3 2.06% 2.43% 18.27% 22.63% 38.03%
17k 1967 43.01% 51.50% 4.02% 1.37% 10.5 15.5 45.9 2.75% 2.59% 19.52% 24.10% 38.38%
16k 2665 45.18% 49.76% 3.53% 1.50% 11.5 14.9 44.2 2.81% 2.59% 19.59% 23.30% 37.79%
15k 3743 44.86% 50.07% 3.79% 1.26% 10.5 14.2 40.5 3.10% 2.73% 20.49% 24.39% 38.10%
14k 4652 44.65% 50.43% 3.65% 1.23% 9.5 13.5 36.5 2.90% 2.54% 21.22% 25.69% 40.11%
13k 5826 44.76% 50.22% 3.54% 1.41% 9.5 12.7 36.5 3.18% 3.02% 22.59% 26.98% 41.14%
12k 7318 43.51% 51.97% 3.39% 1.13% 8.5 12.3 36.4 3.17% 3.46% 24.54% 28.72% 42.77%
11k 9351 43.10% 52.21% 3.60% 1.09% 8.5 12.7 37.5 3.33% 2.98% 23.94% 28.82% 42.15%
10k 11268 45.01% 49.91% 4.22% 0.86% 8.5 12.5 37.5 3.34% 3.01% 23.21% 27.56% 40.83%
9k 12956 46.18% 47.51% 5.54% 0.77% 8.5 11.9 34.5 3.17% 3.16% 23.08% 27.22% 39.62%
8k 12617 47.02% 45.45% 6.82% 0.71% 7.5 11.0 34.5 3.43% 3.23% 24.36% 28.21% 38.85%
7k 13303 48.74% 42.98% 7.71% 0.58% 7.5 10.2 29.7 3.50% 3.09% 23.63% 27.49% 37.74%
6k 14930 50.74% 41.19% 7.76% 0.30% 6.5 9.4 28.5 3.51% 3.45% 24.08% 27.78% 36.85%
5k 11020 52.57% 38.24% 8.88% 0.30% 5.5 8.7 25.5 3.77% 3.45% 23.74% 27.31% 34.96%
4k 7892 48.34% 40.41% 11.14% 0.11% 5.5 8.0 23.5 4.13% 3.71% 26.65% 30.11% 37.58%
3k 6416 46.31% 37.87% 15.65% 0.17% 5.5 7.2 19.5 4.15% 4.25% 25.36% 28.97% 36.16%
2k 3390 49.09% 37.49% 13.27% 0.15% 4.5 6.3 17.5 5.04% 4.01% 26.73% 30.29% 36.22%
1k 1936 54.49% 37.04% 8.37% 0.10% 4.5 5.2 13.5 4.96% 5.22% 29.60% 32.75% 36.52%
1d 863 44.84% 44.15% 10.89% 0.12% 3.5 4.9 12.5 6.37% 6.14% 35.69% 39.17% 43.45%
2d 291 46.05% 44.33% 9.62% 0.00% 3.5 5.1 13.1 7.90% 6.53% 34.71% 38.14% 43.64%
3d 117 46.15% 41.03% 12.82% 0.00% 3.5 5.2 15.5 10.26% 7.69% 30.77% 34.19% 41.03%

Points to notice:

  1. As getting good increases the point difference on average decreases but not necessarily because the games are closer but probably because people know better when to resign.
  2. Timeout rate increases. Could be because more people play blitz or use more of their time.
  3. Disconnection rate decreases. Thus, if you have problems with disconnecting opponents → get good.
  4. Almost 10% disconnection rate for 19x19 25k category.
  5. 95 percentile scores are useful in terms of when to resign. Apparently even 1ds happily score 40-points losses.
8 Likes

Problem with timeout is it can be used as a way to resign for some games. Taking in account if the game was in a lost position could give a better view (but still not satisfactory).

Not so surprising statistic results but it’s good to have them somewhere.

1 Like

Unless those games are something like 15s blitz, generally it is a bannable offence to repeatedly lose by timeout on our server.

2 Likes

Nice but that is far to put it away completly. As you said it has to show like something obvious and systematic (and my guess reported too) looks to be difficult in casual cases to distinguish between a timeout with or without intention because you keep thinking or because unfriendly you don’t care to finish your game.

Who is going to report a game he won in a 30s byo yomi timeout? Sometimes when it’s so exagerated maybe? Especially in blitz but not only, the boundary between resigning and losing on timeout may be less clear as induced by the statistic

5-(strong players finishing lost games with good margin) Could be because more aware that resigning near the end is somewhat unpolite.

Is that so? Interesting point :slightly_smiling_face: My usual practice is to resign as soon as I decide that it’s a hopeless position (what a terrible uncultured 6 kyu Brit I am :laughing:) But are you saying that this is considered poor form if endgame is already half done? :thinking: If so, how close to the end is too close (roughly)? And is this limited to certain national cultures only (if so, which) or is it general across all cultures? Thanks for the observation :+1:

I dont think its impolite to resign when end-game half done, but resigning when theres only passes remaining (or just couple moves before) is a bit rude after taking all the way into that point. But i guess this goes hand-to-hand with players strength and their ability to judge/count the board position - looks like at 10k people start actually counting in their games xD

No-one, theres nothing wrong or unusual about losing on time when you’re on byo-yomi. And on blitz its indeed quite normal way of losing since the “are you sure” pop-up takes around 1-2 seconds go thru.
But letting your time just run as a form of rage-quitting is not allowed.

5 Likes

I think this came first from the Japanese approach of politeness in go, when they teached westerners a few decades ago. You resign following some desaster or at least too big advantage taken in the opening, the middle game or early big yose, but when you start the smaller yose it’s a bit too late to deny the scoring.

Maybe because unless stupidity the game is decided and if you continue to play but finally resign you are considering your opponent a bit stupid.
That can be extended to eastern Asia countries where in their culture the “losing the face” idea is preponderant (yours or the one of your opponent)

3 Likes