What will be the score of knight odds match vs. FM Bolzoni?

[lkaufman]

Just did a quick run on game 1 at 3s/pos, sf6 at 2 threads.

Code: Select all

Move Error Evaluator v9.5

Moves Error calculation based from Stockfish 6 64 POPCNT and players from
Event: Knightoddtest, after round 1.0

Intel(R) Core(TM) i7-2600K CPU @ 3.40GHz
Analyzer id name Stockfish 6 64 POPCNT
Hash 128 MB, Threads 2, MoveTime 3.00 sec/pos
Analyzer error margin: 10 centipawn

 Rank                       Players  AveDepth  TotalPos  AveError  DECnt  E300CP  E200CP  E100CP  E50CP  E20CP  EECnt   EERate    W    L    D  Games  Score  SRate

    1             Komodo 9.1 64-bit        28        62      3.79      3       0       0       0      0      1      6    9.68%    0    0    1      1    0.5   50.0%
    2        BikJump v2.01 (64-bit)        29        62      6.55      2       0       0       0      2      6      1    1.61%    0    0    1      1    0.5   50.0%


Legend:
Rank     - ranking based on low average error
AveDepth - average depth reached by the Analyzer in analyzing those positions
AveError - average difference between the analyzer and player score, in centipawn (1p = 100)
DECnt    - count of positions where player move score is below zero and analyzer move score
         - is better than player move score. This is called Dangerous Error Counts
E300CP   - count of positions where player has blundered by 3 or more pawns
E200CP   - count of positions where player has blundered by 2 to 2.99 pawns
E100CP   - count of positions where player has blundered by 1 to 1.99 pawns
E50CP    - count of positions where player has blundered by 0.5 to 0.99 pawn
E20CP    - count of positions where player has blundered by 0.2 to 0.49 pawn
EECnt    - count of positions where player move score is better than analyzer move score

Notes:
1. Analysis of positions start at move 12.
2. If best score of Analyzer is already 6 pawns and above and the score of player move
   is also 6 pawns and above then error for this position is set to zero. The more cases
   like this only proves that that player is able to maintain its advantage and this
   will minimize its average error at the end.
   On the other hand if the best score of Analyzer is already below minus 6 pawns, no further
   analysis will be done, and this position will not be included in error calculation.

[...]

The 6 pos with 20cp to 49cp from BikJump.

Code: Select all

2kr3r/2p1qppp/1np1p3/4P3/8/P2bPP1P/1B4P1/R1R1Q1K1 b - - fmvn 1; sm "d3b5"; c3 "Result 1/2-1/2"; c2 "Komodo 9.1 64-bit / BikJump v2.01 (64-bit)"; c1 "Round 1"; c0 "Event Knight odd test"; hmvc 0;
2kr3r/2p1qppp/1np1p3/1b2P3/3B4/P3PP1P/6P1/R1R1Q1K1 b - - fmvn 1; sm "b6a4"; c3 "Result 1/2-1/2"; c2 "Komodo 9.1 64-bit / BikJump v2.01 (64-bit)"; c1 "Round 1"; c0 "Event Knight odd test"; hmvc 0;
2k5/2p2ppp/2p1p3/4r3/4P1P1/2R2PKP/8/8 b - - fmvn 1; sm "c8b7"; c3 "Result 1/2-1/2"; c2 "Komodo 9.1 64-bit / BikJump v2.01 (64-bit)"; c1 "Round 1"; c0 "Event Knight odd test"; hmvc 0;
8/1kp2ppp/2p1p3/4r3/4P1P1/3R1PKP/8/8 b - - fmvn 1; sm "c6c5"; c3 "Result 1/2-1/2"; c2 "Komodo 9.1 64-bit / BikJump v2.01 (64-bit)"; c1 "Round 1"; c0 "Event Knight odd test"; hmvc 0;
8/1kpR1ppp/4p3/4r3/2p1P1P1/5P1P/5K2/8 b - - fmvn 1; sm "f7f5"; c3 "Result 1/2-1/2"; c2 "Komodo 9.1 64-bit / BikJump v2.01 (64-bit)"; c1 "Round 1"; c0 "Event Knight odd test"; hmvc 0;
8/1kp5/4p3/2r5/4P1R1/2p4P/2K5/8 b - - fmvn 1; sm "c7c6"; c3 "Result 1/2-1/2"; c2 "Komodo 9.1 64-bit / BikJump v2.01 (64-bit)"; c1 "Round 1"; c0 "Event Knight odd test"; hmvc 0;

I don't have much time to run this at longer time at this time, but I uploaded the tool and epd each round from that pgn. Later I will run this on longer time as I also try to improve the tool.

Estimated max analysis time = analysis time/pos x num_pos x 3. With separated epd by round, and with more cores, this can be analyzed in parallel, just create other folders, copy the tool and epd and run.

Sample run. Just double click the exe file, be sure your engine is in the same dir with the tool. Some files will be generated, the log is in overwrite, but the summary is in append mode. Blunder epd are also saved, and also the sqlite database.

Code: Select all

Move Error Evaluator v9.5

[1] Enter uci engine filename? sf6.exe
[2] Enter epd filename? komo-bik-r1.epd
[3] Enter Hash size in MB? 128
[4] Enter number of Threads? 2
[5] Enter analysis time/pos in sec? 3

Analyzer error margin is a value in cp used to decide whether to
save the position if player score is better than Analyzer score.
If this margin is low say zero then all positions will be saved.
It is recommended to set this margin to high say 10cp if analysis
time given to the analyzer is low say below 30s. The HW specs,
Analyzer engine strength and its settings could also be a factor.

[6] Enter Analyzer error margin in centipawn? 10

Creating player and pairing list ...
Done creating player and pairing list
Time elapsed 1: 0.3 sec

Player: Komodo 9.1 64-bit
Evaluate new position (62) of Komodo 9.1 64-bit

player               : Komodo 9.1 64-bit
total error          : 235 cp
total considered pos : 62
average error        : 4 cp

Player: BikJump v2.01 (64-bit)
Evaluate new position (62) of BikJump v2.01 (64-bit)

player               : BikJump v2.01 (64-bit)
total error          : 406 cp
total considered pos : 62
average error        : 7 cp

Time elapsed 2: 695.4 sec

Done!!
press enter key to exit

Download:
http://www.mediafire.com/download/kofa7 ... E_V9.5.rar

Someone might be interested to analyze the old match rybka-meyer (at longer time and with more cpu power). Here is the epd in one file. It is better to run this by round so the summary will be separated. There is round in epd file, so it is easy to create r1.epd to r4.epd, then run with the tool. Don't use longer time/pos at first, try it first with say 1s/pos, see if there is problem. I only tested these in short time.

Download:
http://www.mediafire.com/download/s15m1 ... -meyer.rar

I guess the 57% score for White at this level is about par. So this line is more playable than generally believed for Black. After this match, do you like the ...Qd6 defense more, less, or about the same as before?

[lkaufman]

So far the median of the votes is for an even score. It looks like we picked the handicap that has the most suspense for this match!

[JJJ]

Yeap.
Did you find higher rated candidate for the next matches ?

[lkaufman]

Yeap.
Did you find higher rated candidate for the next matches ?

We have two GMs lined up for the two pawn handicap, and if we do win this knight odds match I can get John Meyer (FM,2270) back for a knight odds match, since he beat Rybka four to zero.

[yanquis1972]

i voted 6/6, figured it would be highest or maybe next to 5/6...i wasn't exactly wrong, but my vote tied it with 2/6! will definitely want to follow this match

[Uri Blass]

I think that it may be interesting to see komodo play for fide rating when komodo play without knight b1 at time control of 90+30 that is the normal time control(slower than 45+15)

What is going to be komodo's fide rating in this case?

A match against a single human can give misleading result because it is possible that the human is relatively better in winning won positions or relatively worse in winning won positions.

[Uri Blass]

If fide does not agree to allow Komodo to play for fide rating in these conditions then maybe it is possible to convince local organizations to allow komodo to play for local rating without knight b1 for some money
or maybe with smaller handicap.

Chess computers played in the israeli league in year 2000 and maybe it is possible to convince the israeli chess organization to allow computers again and this time without knight b1 or without pawns c2 f2 but before I ask them the question is if the komodo team is interested to do it.

I guess that you need to pay the israeli organization some money for it
when part of it can come from the komodo team if they are interested in it and part of it can come from the team that want to use komodo.

Time control for first league in israel is 90 minutes/40 moves+30 minutes+30 seconds for every move.

I think that knight b1 is too much for the first league in israel but c2 and f2 can be good enough at least for last board when the average fide rating is something near 2300.

[Vinvin]

Yeap.
Did you find higher rated candidate for the next matches ?

We have two GMs lined up for the two pawn handicap...

Is this involved the f pawn ? There will be a draw odd too ?

[lkaufman]

If fide does not agree to allow Komodo to play for fide rating in these conditions then maybe it is possible to convince local organizations to allow komodo to play for local rating without knight b1 for some money
or maybe with smaller handicap.

Chess computers played in the israeli league in year 2000 and maybe it is possible to convince the israeli chess organization to allow computers again and this time without knight b1 or without pawns c2 f2 but before I ask them the question is if the komodo team is interested to do it.

I guess that you need to pay the israeli organization some money for it
when part of it can come from the komodo team if they are interested in it and part of it can come from the team that want to use komodo.

Time control for first league in israel is 90 minutes/40 moves+30 minutes+30 seconds for every move.

I think that knight b1 is too much for the first league in israel but c2 and f2 can be good enough at least for last board when the average fide rating is something near 2300.

We would be glad to have Komodo participate if it's allowed, but we're only paying 25 euros per win in these knight-odds matches so that's probably not going to be enough to interest them. But we'll probably play more than one match at knight odds so as to minimize the chance that one particular opponent was atypical.

[lkaufman]

Yeap.
Did you find higher rated candidate for the next matches ?

We have two GMs lined up for the two pawn handicap...

Is this involved the f pawn ? There will be a draw odd too ?

The two pawns are f2 and c2. No draw odds, that's only for knight odds, which is roughly twice as big a handicap as this. In the opening, a piece is worth at least four pawns, not three.