Please turn on javascript in your browser to play chess.
Only Chess Forum

Only Chess Forum

  1. 02 May '11 18:29
    (The subject of move match-up rates came up in another thread, and it caught my interest.)

    I just ran my match-up rates, and I'm shocked. My 1st choice m/u is 50%, and my top 3 m/u is 85%. I expected it to be much lower. (My USCF rating is in the 1200s, for God's sake.)

    I haven't played any RHP games in a while, and I only have 26 total.

    I excluded the opening phase of the games where I knew I was still in theory. (Sadly, usually only 4 or 5 moves deep at most. I think I may have played 1 or 2 games where I actually consulted opening books.) I did run the m/u rates to the end of all games.

    To all you guys that follow these things - Do these numbers sound about right for someone with an RHP rating of about 1680 that takes a lot of time on his moves? Or will I now be labeled a scummy cheater? (For the record, I have never cheated in chess, either here or elsewhere. Scout's honor!)
  2. Standard member nimzo5
    Ronin
    02 May '11 18:47
    So funny that you did that, because after reading the other thread I was bored and ran the 8 games that I have that qualify min rating 2000.

    my matchup rates were
    62% 82% using 2008 big base
    58% 78% using my openings analysis and my database of 6.5 mil games. 3 of my 8 games there was no difference in my prep and a generic database.

    Also interesting is that in my games my opponents matchups were
    65% 83%

    My record in these games was +3 -3 =2 were in two games I had massive blunders, one was a missed win (+9) that I blundered down to +0 (i won 30 moves later.. ughh.) and the biggest blunder I resigned thinking I had blundered a minor piece in an ending when in fact I had a tactical resource I overlooked.
  3. 02 May '11 19:09 / 2 edits
    This is just a question of not understanding the match rate methodology.
    Anyone can match highly in a single or few cherry-picked games.

    Take Kramnik's match rate from game 10 in the controversial 2006 FIDE World Championships:

    Fritz 11 @ 30 seconds per move
    4x AMD Phenom 2.30Ghz
    4GB DDR2 RAM
    Hash Table 512MB
    Database used www.chesslive.de

    [Event "World Championship"]
    [Site "Elista"]
    [Date "2006.09.23"]
    [Round "10"]
    [White "Kramnik, Vladimir"]
    [Black "Topalov, Veselin"]
    [Result "1-0"]
    [ECO "E08"]
    [PlyCount "85"]
    [EventDate "2006.??.??"]

    1. d4 Nf6 2. c4 e6 3. Nf3 d5 4. g3 Bb4+ 5. Bd2 Be7 6. Bg2 O-O 7. O-O c6 8. Bf4
    Nbd7 9. Qc2 a5 10. Rd1 Nh5 11. Bc1 b5 12. cxd5 {Takes game out of book; 2nd}
    cxd5 {1st} 13. e4 {1st} dxe4 {Not in top 3} 14. Qxe4 {1st} Rb8 {3rd} 15. Qe2 {
    1st} Nhf6 {2nd} 16. Bf4 {1st} Rb6 {1st} 17. Ne5 {1st} Nd5 {1st} 18. Bxd5 {1st}
    exd5 {1st} 19. Nc3 {1st} Nf6 {Not in top 3} 20. Nxb5 {1st} Ba6 {1st} 21. a4 {
    1st} Ne4 {2nd} 22. Rdc1 {1st} Qe8 {1st} 23. Rc7 {Not in top 3} Bd8 {1st} 24.
    Ra7 {3rd} f6 {Not in top 3} 25. Nd7 {1st} Rf7 {1st} 26. Nxb6 {1st} Rxa7 {1st}
    27. Nxd5 {1st} Rd7 {1st} 28. Ndc3 {1st} Rxd4 {Not in top 3} 29. Re1 {2nd} f5 {
    1st} 30. Qc2 {1st} Rb4 {1st} 31. Nd5 {2nd} Rxb5 {1st} 32. axb5 {1st} Qxb5 {1st}
    33. Nc7 {3rd} Qc4 {2nd} 34. Qd1 {2nd} Bxc7 {1st} 35. Qd7 {1st} h6 {1st} 36.
    Qxc7 {2nd} Qb4 {3rd} 37. Qb8+ {1st} Qxb8 {1st} 38. Bxb8 {1st} Nd2 {3rd} 39. Ra1
    {1st} g5 {Not in top 3} 40. f4 {1st} Nb3 {1st} 41. Ra3 {2nd} Bc4 {1st} 42. Bc7
    {2nd} g4 {Not in top 3} 43. Bxa5 {1st} 1-0

    Result:
    White: Kramnik
    Top 1 Match: 22/32 (68,8% )
    Top 2 Match: 29/32 (90,6% )
    Top 3 Match: 31/32 (96,9% )

    Black: Topalov
    Top 1 Match: 19/31 (61,3% )
    Top 2 Match: 22/31 (71,0% )
    Top 3 Match: 25/31 (80,6% )

    And yet Kramnik couldn't sustain such engine-like play throughout the matches - here are his total results from the 2006 WC. I also include Topalov's results & he was accused of consulting an engine at various stages during this match:

    Kramnik (14 games)
    Top 1 Match: 306/552 (55,4% )
    Top 2 Match: 417/552 (75,5% )
    Top 3 Match: 461/552 (83,5% )

    Topalov (14 games)
    Top 1 Match: 309/555 (55,7% )
    Top 2 Match: 417/555 (75,1% )
    Top 3 Match: 447/555 (80,5% )

    If you can achieve match rates in 20 objectively chosen games vs quality opposition (let's say RHP 2000+ rated) which all have 20 or more non-theory moves which surpass the benchmark thresholds, then you play more engine-like chess than any unassisted player so far tested.

    Either that or you're a cheating ^%£*&^% 😉
  4. 02 May '11 19:15
    Originally posted by nimzo5
    So funny that you did that, because after reading the other thread I was bored and ran the 8 games that I have that qualify min rating 2000.

    my matchup rates were
    62% 82% using 2008 big base
    58% 78% using my openings analysis and my database of 6.5 mil games. 3 of my 8 games there was no difference in my prep and a generic database.

    Also interesting ...[text shortened]... I had blundered a minor piece in an ending when in fact I had a tactical resource I overlooked.
    nimzo5
    If you like I can analyse 20 of your games which meet the basic criteria.
    The games I'll select will be the 20 most recent vs 2000+ rateds which all have 20 or more non-www.chesslive.de moves.
    I can even publish the end results here if you like. I'm certain that they will be well under the thresholds, since you don't consult an engine in any of your games 🙂
  5. 02 May '11 19:18
    Originally posted by nimzo5
    So funny that you did that, because after reading the other thread I was bored and ran the 8 games that I have that qualify min rating 2000.

    my matchup rates were
    62% 82% using 2008 big base
    58% 78% using my openings analysis and my database of 6.5 mil games. 3 of my 8 games there was no difference in my prep and a generic database.

    Also interesting ...[text shortened]... I had blundered a minor piece in an ending when in fact I had a tactical resource I overlooked.
    Which games are these?
    Can you please post the game id's & I'll take a look...
  6. Standard member nimzo5
    Ronin
    02 May '11 19:45 / 2 edits
    I have only found 8 so far that I have found that fit your criteria, but I am still looking.
  7. Standard member nimzo5
    Ronin
    02 May '11 20:34
    Originally posted by Zygalski
    nimzo5
    If you like I can analyse 20 of your games which meet the basic criteria.
    The games I'll select will be the 20 most recent vs 2000+ rateds which all have 20 or more non-www.chesslive.de moves.
    I can even publish the end results here if you like. I'm certain that they will be well under the thresholds, since you don't consult an engine in any of your games 🙂
    I would feel a lot better about your method if you used something a tad more advanced than a generic chessbase database.

    My interest in running through my matchup rates was too see just how far off my own combination of opening prep/databases effects my overall rates. So far it is more than I would have thought (roughly 4-5% on first move matchups). Likewise I found my matchup rates skewed in certain openings Najdorf very high, Queens gambit very low.. perhaps high matchup rate in closed positions is more meaningful?

    Anyway, a lot of things to think about and no time at the moment.
  8. 03 May '11 11:45 / 1 edit
    Hallo,

    can somebody pass on a link or quickly describe how to do this analysis in a convenient manner? Is there already a program online, where you just past in a PGN for analysis or will it be hand-done always?

    I am curious, but also very lazy and don't know, if it makes sense when playing around 1700...

    t

    edit: madrook, got some advice to save time?
  9. 03 May '11 12:09
  10. 03 May '11 13:10
    thanks, this thread really is a good guide...

    but one hour for one game... i feel my laziness...
  11. 03 May '11 14:01 / 3 edits
    Originally posted by tharkesh
    Hallo,

    can somebody pass on a link or quickly describe how to do this analysis in a convenient manner? Is there already a program online, where you just past in a PGN for analysis or will it be hand-done always?

    I am curious, but also very lazy and don't know, if it makes sense when playing around 1700...

    t

    edit: madrook, got some advice to save time?
    I wasn't aware of the thread that GP posted. Looks like good advice. A good number of my games were against weaker players, which may have skewed my match-up rate higher than it should be. (That's a relief; My m/u rate shouldn't be that high.)

    I was using a demo version of a program called ChessAnalyze ver 2.6, available here:

    http://www.chess.com/download/view/chessanalyse-25

    Unfortunately, it ain't free if you want all the features. But the demo is still useful (but stops working after 30 days). (The last page of the pdf manual lists the restrictions of the demo version.)

    One thing that I just discovered that I don't understand about the program - There's a setting named "Analyze actual move" that seems to give different results, depending on whether it's checked or not - And I don't yet fully understand its function. (I kind of understand it, but not really. If that makes any sense. I've read the explanation in the pdf manual, and it's still clear as mud to me.)

    I may manually analyze 1 or two of my games and compare the results to the ChessAnalyze results.
  12. 03 May '11 14:39
    Originally posted by Mad Rook
    (The subject of move match-up rates came up in another thread, and it caught my interest.)

    I just ran my match-up rates, and I'm shocked. My 1st choice m/u is 50%, and my top 3 m/u is 85%. I expected it to be much lower. (My USCF rating is in the 1200s, for God's sake.)

    I haven't played any RHP games in a while, and I only have 26 total.

    I excluded ...[text shortened]... r? (For the record, I have never cheated in chess, either here or elsewhere. Scout's honor!)
    Can you post the links to the games you analysed?
    I'd be interested to look at them.
  13. 03 May '11 15:09
    Originally posted by Zygalski
    Can you post the links to the games you analysed?
    I'd be interested to look at them.
    They were basically all of my games (I've only played 26 in total). I didn't use ID 3301368, as it was a two-move timeout - I guess my opponent didn't want to play the Petrov. But here they are:

    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=8079409
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=8082056
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=8079406
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=8079261
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=8079411
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=7196347
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=7187587
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=7171482
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=7191803
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=7159276
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=7174416
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=7156710
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=7171550
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=7156938
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=7159447
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=7158490
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=3529887
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=3508454
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=3365905
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=3298754
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=3296363
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=3298642
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=3314049
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=3305998
    http://www.redhotpawn.com/gameanalysis/boardhistory.php?gameid=3301141

    Like I said, a good number of these games were against weaker and/or provisional players, which may have skewed my m/u rate higher.
  14. 03 May '11 15:28
    Ok, I'll batch analyse them & discount the games with less than 20 non-database moves from the end results.
    As you say, playing games with a few moves once out of book vs weaker opposition, or simply playing a drawn-out easily won game can skew results, but it will be interesting to see what happens...
  15. 04 May '11 00:51 / 1 edit
    Originally posted by Zygalski
    Fritz 11 @ 30 seconds per move
    4x AMD Phenom 2.30Ghz
    4GB DDR2 RAM
    Hash Table 512MB
    Database used www.chesslive.de
    I have noticed in a couple of posts that the some example analyses of match-up rates involves Fritz @ 30 seconds per move.

    Yet, if a cheat has his Fritz engine run say one hour per move (instead of only 30 seconds), could not that distort a matchup analysis based on 30 seconds per move.

    In other words, doesn't it occur with at least a little regularity that a low choice move (e.g., 4th or 5th) at 30 seconds may rise to a high choice move (e.g., 1st or 2nd) at one hour analysis, or vice versa.

    It seems that a conclusion on matchup rates could be significantly dependent on engine analysis time, or depth, or number of leaves (if I am saying that correctly). How does the analyzer determine what engine analysis time/depth the engine cheater is using.