go19.ch.12.1.1 predictions Go forum

23 replies. Last post: 2015-09-13

Reply to this topic Return to forum

go19.ch.12.1.1 predictions
  • RoRoRo the Bot at 2014-05-14

    Tournament: go19.ch.12.1.1
    I’ve been requested to follow go19.ch.12.1.1 by RoRoRo the Bot
    For info on how these predictions are calculated see my predictions web page .

    • Already completed matches:

    Games completed: 0/36

    N Player Rating 1 2 3 4 5 6 7 8 9 Score Son p(win) p(draw) p(down)
    1 LEESD 2490 - _ _ _ _ _ _ _ _ 0 0 10.97% 1.32% 31.96%
    2 plici 2525 _ - _ _ _ _ _ _ _ 0 0 17.19% 1.80% 23.10%
    3 Xaver Gubas 2530 _ _ - _ _ _ _ _ _ 0 0 18.33% 1.76% 22.05%
    4 gamesorry 2540 _ _ _ - _ _ _ _ _ 0 0 20.78% 1.87% 19.68%
    5 Crelo 2530 _ _ _ _ - _ _ _ _ 0 0 18.27% 1.87% 22.78%
    6 mafi 2435 _ _ _ _ _ - _ _ _ 0 0 4.85% 0.86% 53.40%
    7 sspring 2400 _ _ _ _ _ _ - _ _ 0 0 2.62% 0.45% 66.55%
    8 mirror_man2 2370 _ _ _ _ _ _ _ - _ 0 0 1.57% 0.34% 75.06%
    9 erratic 2325 _ _ _ _ _ _ _ _ - 0 0 0.66% 0.16% 85.41%

    p(down) is the probability of being in the places 6-9 and thus being demoted

  • RoRoRo the Bot at 2014-05-28

    Tournament: go19.ch.12.1.1 Newly completed matches: * Xaver Gubas 2 – 0 LEESD Games completed: 1/36 table{background:#ffe700;border:1}. {background:#ff8000}. |_. N |_. Player |_. Rating |_. 1 |_. 2 |_. 3 |_. 4 |_. 5 |_. 6 |_. 7 |_. 8 |_. 9 |_. Score |_. Son |_. p(win) |_. p(draw) |_. p(down) | | 1 | Xaver Gubas | 2530 | - | _ | _ | 2 | _ | _ | _ | _ | _ | 2 | 0 | 26.29% | 2.19% | 11.63% | | 2 | erratic | 2325 | _ | - | _ | _ | _ | _ | _ | _ | _ | 0 | 0 | 0.61% | 0.17% | 82.92% | | 3 | sspring | 2405 | _ | _ | - | _ | _ | _ | _ | _ | _ | 0 | 0 | 2.86% | 0.51% | 61.35% | | 4 | LEESD | 2480 | 0 | _ | _ | - | _ | _ | _ | _ | _ | 0 | 0 | 3.66% | 0.76% | 47.86% | | 5 | mafi | 2425 | _ | _ | _ | _ | - | _ | _ | _ | _ | 0 | 0 | 4.06% | 0.70% | 55.39% | | 6 | mirror_man2 | 2370 | _ | _ | _ | _ | _ | - | _ | _ | _ | 0 | 0 | 1.48% | 0.29% | 73.42% | | 7 | plici | 2525 | _ | _ | _ | _ | _ | _ | - | _ | _ | 0 | 0 | 16.98% | 1.68% | 24.38% | | 8 | gamesorry | 2540 | _ | _ | _ | _ | _ | _ | _ | - | _ | 0 | 0 | 20.42% | 1.91% | 20.79% | | 9 | Crelo | 2535 | _ | _ | _ | _ | _ | _ | _ | _ | - | 0 | 0 | 19.00% | 1.89% | 22.25% | p(down) is the probability of being in the places 6-9 and thus being demoted

  • RoRoRo the Bot at 2014-05-30

    Tournament: go19.ch.12.1.1 Newly completed matches: * Xaver Gubas 0 – 2 sspring Games completed: 2/36 table{background:#ffe700;border:1}. {background:#ff8000}. |_. N |_. Player |_. Rating |_. 1 |_. 2 |_. 3 |_. 4 |_. 5 |_. 6 |_. 7 |_. 8 |_. 9 |_. Score |_. Son |_. p(win) |_. p(draw) |_. p(down) | | 1 | sspring | 2425 | - | 2 | _ | _ | _ | _ | _ | _ | _ | 2 | 4 | 8.75% | 1.26% | 35.46% | | 2 | Xaver Gubas | 2510 | 0 | - | _ | 2 | _ | _ | _ | _ | _ | 2 | 0 | 11.50% | 1.58% | 27.11% | | 3 | erratic | 2325 | _ | _ | - | _ | _ | _ | _ | _ | _ | 0 | 0 | 0.75% | 0.20% | 84.16% | | 4 | LEESD | 2480 | _ | 0 | _ | - | _ | _ | _ | _ | _ | 0 | 0 | 5.04% | 0.89% | 49.44% | | 5 | mirror_man2 | 2370 | _ | _ | _ | _ | - | _ | _ | _ | _ | 0 | 0 | 1.79% | 0.37% | 74.08% | | 6 | mafi | 2425 | _ | _ | _ | _ | _ | - | _ | _ | _ | 0 | 0 | 4.53% | 0.73% | 58.19% | | 7 | plici | 2525 | _ | _ | _ | _ | _ | _ | - | _ | _ | 0 | 0 | 18.79% | 1.99% | 25.49% | | 8 | gamesorry | 2540 | _ | _ | _ | _ | _ | _ | _ | - | _ | 0 | 0 | 22.56% | 1.98% | 22.27% | | 9 | Crelo | 2535 | _ | _ | _ | _ | _ | _ | _ | _ | - | 0 | 0 | 21.23% | 2.06% | 23.80% | p(down) is the probability of being in the places 6-9 and thus being demoted

  • RoRoRo the Bot at 2014-07-23

    Tournament: go19.ch.12.1.1 Newly completed matches: * sspring 2 – 0 mafi * Xaver Gubas 2 – 0 Crelo * mafi 0 – 2 LEESD * Crelo 2 – 0 mafi * mafi 2 – 0 Xaver Gubas * LEESD 0 – 2 sspring * sspring 0 – 2 Crelo Games completed: 9/36 table{background:#ffe700;border:1}. {background:#ff8000}. |_. N |_. Player |_. Rating |_. 1 |_. 2 |_. 3 |_. 4 |_. 5 |_. 6 |_. 7 |_. 8 |_. 9 |_. Score |_. Son |_. p(win) |_. p(draw) |_. p(down) | | 1 | sspring | 2475 | - | 0 | 2 | 2 | 2 | _ | _ | _ | _ | 6 | 16 | 19.32% | 2.77% | 9.15% | | 2 | Crelo | 2485 | 2 | - | 0 | 2 | _ | _ | _ | _ | _ | 4 | 16 | 13.56% | 1.81% | 19.51% | | 3 | Xaver Gubas | 2520 | 0 | 2 | - | 0 | 2 | _ | _ | _ | _ | 4 | 12 | 8.58% | 1.70% | 22.35% | | 4 | mafi | 2405 | 0 | 0 | 2 | - | 0 | _ | _ | _ | _ | 2 | 8 | 0.15% | 0.13% | 84.30% | | 5 | LEESD | 2475 | 0 | _ | 0 | 2 | - | _ | _ | _ | _ | 2 | 4 | 1.94% | 0.58% | 61.85% | | 6 | plici | 2540 | _ | _ | _ | _ | _ | - | _ | _ | _ | 0 | 0 | 22.48% | 1.98% | 23.14% | | 7 | mirror_man2 | 2395 | _ | _ | _ | _ | _ | _ | - | _ | _ | 0 | 0 | 2.63% | 0.44% | 70.94% | | 8 | gamesorry | 2550 | _ | _ | _ | _ | _ | _ | _ | - | _ | 0 | 0 | 25.22% | 2.10% | 21.57% | | 9 | erratic | 2325 | _ | _ | _ | _ | _ | _ | _ | _ | - | 0 | 0 | 0.69% | 0.22% | 87.19% | p(down) is the probability of being in the places 6-9 and thus being demoted

  • erratic at 2014-07-23

    I am the underdog.

    Great!

    This site needs betting functionality urgently ;-)

  • erratic at 2014-07-27

    Trying to clean the markdown and sorting by probability of being demoted the table stands_

    |aaa|aaa|aaa|aaa|aaa|aaa|aaa|aaa|aaa|aaa|aaa|aaa|aaa|aaa|aaa|aaa|aaa|
    |-|-|-|-|-|-|-|-|-|-|-|-|-|-|-|-|-|
    | 1 | sspring | 2475 | - | 0 | 2 | 2 | 2 | _ | _ | _ | _ | 6 | 16 | 19.32% | 2.77% | 9.15% |
    | 2 | Crelo | 2485 | 2 | - | 0 | 2 | _ | _ | _ | _ | _ | 4 | 16 | 13.56% | 1.81% | 19.51% |
    | 8 | gamesorry | 2550 | _ | _ | _ | _ | _ | _ | _ | - | _ | 0 | 0 | 25.22% | 2.10% | 21.57% |
    | 3 | Xaver Gubas | 2520 | 0 | 2 | - | 0 | 2 | _ | _ | _ | _ | 4 | 12 | 8.58% | 1.70% | 22.35% |
    | 6 | plici | 2540 | _ | _ | _ | _ | _ | - | _ | _ | _ | 0 | 0 | 22.48% | 1.98% | 23.14% |
    | 5 | LEESD | 2475 | 0 | _ | 0 | 2 | - | _ | _ | _ | _ | 2 | 4 | 1.94% | 0.58% | 61.85% |
    | 7 | mirror_man2 | 2395 | _ | _ | _ | _ | _ | _ | - | _ | _ | 0 | 0 | 2.63% | 0.44% | 70.94% |
    | 4 | mafi | 2405 | 0 | 0 | 2 | - | 0 | _ | _ | _ | _ | 2 | 8 | 0.15% | 0.13% | 84.30% |
    | 9 | erratic | 2325 | _ | _ | _ | _ | _ | _ | _ | _ | - | 0 | 0 | 0.69% | 0.22% | 87.19% |

    Hope it is more readable

  • erratic at 2014-07-27

    Markdown doesn’t work but it is more readable anyway.

    The algorithm has chosen 4 of us (LEESD. mirror_man2, mafi and myself) as demoted >61% and the others as “saved” <24% :(

    How does it make enormous differences between plici (23%) and mirror_man2 (70.9%) if neither has a finished game and their score is very close?


  • William Fraser at 2014-07-27

    RoRoRo runs the tournament 10,000 times, with each unfinished game having the outcome randomly determined based on the difference in ratings of the two players (the position in the actual game itself is not considered).

    The reported probabilities are then based on what percentage of the simulations end up with a given result.

  • erratic at 2014-07-28

    ???

  • erratic at 2014-07-28

    ???

  • erratic at 2014-07-28

    ???

  • erratic at 2014-07-28

    (chunk 1)

  • erratic at 2014-07-28

    (Try again)

    So the mistake is trusting the rating system. The strongest go player is a 9x9 bot and the 2nd strongest was demoted (7th) from league 1 last tournament.

    http://www.littlegolem.net/jsp/tournament/tournament.jsp?trnid=go19.ch.11.1.1

    The rating system has many known problems (handicap, systematic bias from the starting point, no management of sandbaggers/abandoners, etc.), the worst of all is merging board sizes, which are different games (same rules but different probability of winning). It is not even plausible that go 9x9 has as many levels as go 19x19 which makes the whole merging idea a misconception. If at all, the ratings should be transformed from one board size to another, and finding the correct transformation function is nontrivial. Not to mention some board sizes are mastered by machines and others by humans.

    Also, the difference is way too exaggerated. A steep change from 23% to 61% from one player to the next (first 5 vs. next 4) when the truth is much more equilibrated. Maybe less weight to the rating should improve the prediction and estimating winning probabilities from the players history even more.


    Note: If I am finally demoted, which of course may happen, I will have to eat my words, but still I rebel against the 87% probability of that happening.

    And thanks to RoRoRo! I am happy to have a prediction.

  • erratic at 2014-07-28

    Sorry for the multiple post. (The problem was copy/pasting it from Word. Incompatible formatting?)

  • sspring at 2014-08-02

    Nothing serious.  Just playing for fun, and no awards, right?


    I think the four guys who play extreme slow take it too serious (No time visiting this site seems not the case for them).

  • erratic at 2014-08-09

    Slow? I think in the first 100 moves of a tournament that takes 17 months approx you never play slow enough. 8 games are a lot of games. I enjoy studying the games, that’s why I play in LG. And anyway, my games never take more than 7 or 8 months (which is less than half the tournament time). At some point I either resign or know the position so well that I can answer as soon as my opponent plays.

    For me playing fast as the 4 guys who have finished their “private tournament” is a strange strategy. I only consider that optimal if you are sure you can win those games and therefore have more time for the others. But I have no idea what games I could win, so out of respect for my opponents, I play all the games as slow as I can.

    Playing fast and losing a game is obviously suboptimal.

  • RoRoRo the Bot at 2014-09-13

    Tournament: go19.ch.12.1.1 Newly completed matches: * plici 2 – 0 Xaver Gubas Games completed: 10/36 table{background:#ffe700;border:1}. {background:#ff8000}. |_. N |_. Player |_. Rating |_. 1 |_. 2 |_. 3 |_. 4 |_. 5 |_. 6 |_. 7 |_. 8 |_. 9 |_. Score |_. Son |_. p(win) |_. p(draw) |_. p(down) | | 1 | sspring | 2390 | - | 0 | 2 | 2 | _ | 2 | _ | _ | _ | 6 | 16 | 10.76% | 1.71% | 15.39% | | 2 | Crelo | 2440 | 2 | - | 0 | 2 | _ | _ | _ | _ | _ | 4 | 16 | 10.08% | 1.49% | 24.73% | | 3 | Xaver Gubas | 2510 | 0 | 2 | - | 0 | 0 | 2 | _ | _ | _ | 4 | 12 | 0.79% | 0.65% | 32.42% | | 4 | mafi | 2375 | 0 | 0 | 2 | - | _ | 0 | _ | _ | _ | 2 | 8 | 0.14% | 0.11% | 85.40% | | 5 | plici | 2570 | _ | _ | 2 | _ | - | _ | _ | _ | _ | 2 | 8 | 52.61% | 2.70% | 5.31% | | 6 | LEESD | 2475 | 0 | _ | 0 | 2 | _ | - | _ | _ | _ | 2 | 4 | 3.07% | 0.82% | 54.31% | | 7 | erratic | 2325 | _ | _ | _ | _ | _ | _ | - | _ | _ | 0 | 0 | 1.06% | 0.27% | 80.66% | | 8 | mirror_man2 | 2370 | _ | _ | _ | _ | _ | _ | _ | - | _ | 0 | 0 | 2.47% | 0.47% | 69.87% | | 9 | gamesorry | 2485 | _ | _ | _ | _ | _ | _ | _ | _ | - | 0 | 0 | 14.61% | 1.46% | 31.90% | p(down) is the probability of being in the places 6-9 and thus being demoted

  • erratic at 2014-09-13












    NPlayerRating123456789ScoreSonp(win)p(draw)p(down)
    1sspring2390022261610.76%1.71%15.39%
    2Crelo244020241610.08%1.49%24.73%
    3Xaver Gubas251002-0024120.79%0.65%32.42%
    4mafi23750020280.14%0.11%85.40%
    5plici25702_2852.61%2.70%5.31%
    6LEESD2475002243.07%0.82%54.31%
    7erratic2325_001.06%0.27%80.66%
    8mirror_man22370-002.47%0.47%69.87%
    9gamesorry2485-0014.61%1.46%31.90%


  • erratic at 2015-01-10

    Will this finish with all 9 players tied as first??

    I would really like to see what RoRoRo thinks now. During the last months I though 4 wins meant avoiding demotion and that was my objective. Now I see everybody can make that (realistically, not just in theory). The only finished player, Xaver,  has 4 wins/4loses, mafi is comfortably leading his 2 remaining games to achieve 4wins/4loses. I myself have 3wins/3loses/2anythingcanhappen, LEESD is ok in his 2 remaining games and can still make 4wins/4loses.

    Will plici win more than 4 games (he has the best win/loss ratio) and leave some players with less than 4 wins? If so, 4 wins is no longer safe and the lottery of son will decide who gets demoted.

    Again, this is worth betting.

  • ypercube at 2015-01-10

    @erratic, sspring is also on 3 wins / 1 loss currently and the game between sspring and plici is still open. One of them will reach 4 wins with 3 remaining games.

  • erratic at 2015-01-23

    It is broken!! Crelo has 5 wins.

    The tournament will be won by someone with 5/3 (10 points) or 6/2 (12 points). Still betting on plici.

    There will be less than 4 players under 4 wins, so the demotion cut will be decided by SON among the players with 4wins/4loses. Nobody is safe (except players with more than 4 wins, of course).

  • erratic at 2015-09-13

    An amazing tournament !!!
    Thank you to all opponents. Note that the level was (despite rororo’s predictions) so balanced that just my last game (won by 1.5 points in extremis) lost would have taken me from my third position to the sixth and I would have demoted. But still my 87% of probability of demotion was too high and I finally got the “bronze medal”. Also note that the winner (plici) lost to mirror_man2 who is demoted. Again, extremely hard and equal level for anyone. Not a single weak player.
    Maybe I really was the weakest, but dedicated an insane amount of time to this, so now I need a break. Anyway, I doubt I could make third again, so this is my best result forever.
    I improved a lot playing here and it was really fun.
    Cheers.

  • gamesorry at 2015-09-13

    Amazing tournament indeed :) I happened to be the one with the highest rating when the tournament started (due to a winning streak in the smaller boards then), but only managed to win one game (another win was a timeout) and placed in the last spot finally.
    @erratic You earned the place by your play, and I hope you would come back one day (and I hope I could as well ;-)).

Return to forum

Reply to this topic




Include game board: [game;id:123456] or [game;id:123456;move:20] or [game;id:123456;move:20;title:some text]