Search found 2079 matches

by Michel
Wed Jul 08, 2020 6:51 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Stockfish has included WDL stats in engine output
Replies: 58
Views: 4045

Re: Stockfish has included WDL stats in engine output

Before discussing the SF wdl data one should understand what they actually represents. There are two principal definitions of wdl. Given a particular position one could define it as the empirical probabilities that an engine scores a win,draw,loss against itself when playing from that position. Most...
by Michel
Tue Jul 07, 2020 10:12 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Stockfish has included WDL stats in engine output
Replies: 58
Views: 4045

Re: Stockfish has included WDL stats in engine output

If you want to end up with real win rates, just run a long match with SF against some other engine or against itself and collect the stats. It's not so clear how to objectively measure the win rate of a particular position. One must have some method for introducing variety. The method one uses may ...
by Michel
Mon Jul 06, 2020 8:35 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Throwing out draws to calculate Elo
Replies: 214
Views: 4614

Re: Throwing out draws to calculate Elo

This whole discussion is very strange. Lots of useless words for things which are well understood and well known. LOS is a Bayesian concept. Hence it is not an empirical probability but a degree of belief. Formally LOS is the probability that one engine is stronger than another assuming a uniform pr...
by Michel
Tue Feb 25, 2020 9:12 pm
Forum: Computer Chess Club: General Topics
Topic: Repeating games with switched colors reduces Elo error. All matches should be done like this
Replies: 40
Views: 5314

Re: Repeating games with switched colors reduces Elo error. All matches should be done like this

This only applies to tests with the same opening book for both sides. It makes intuitive sense that the results will be more accurate if player A and player B play both sides of all openings. But I couldn't find any empirical results so I wrote a utility to test it out myself. First, I've compared ...
by Michel
Sat Feb 22, 2020 12:56 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: EPD destruction tests
Replies: 81
Views: 7403

Re: EPD destruction tests

First two are NOT correct. White or black can never have more that 16 pieces counting king. Not 17 or 18. The rules of chess say 16 pieces. 8 pawns, two rooks, bishops and knights, queen and king. The pawns can promote into 8 pieces excluding kings So 16 is it unless you are doing non-chess. For th...
by Michel
Mon Feb 17, 2020 7:55 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: hash collisions
Replies: 370
Views: 35745

Re: hash collisions

This whole discussion is silly. About the only hard requirement of a chess program is that it plays legal moves when given a legal position. So a "bug" in a chess program is mostly an undefined concept. A pv must be sensible? Some people here will argue that a pv is only "cosmetics"... Endgame troll...
by Michel
Mon Feb 03, 2020 11:04 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Stockfish Reverts 5 Recent Patches
Replies: 9
Views: 2846

Re: Stockfish Reverts 5 Recent Patches

Michel wrote:
Sun Feb 02, 2020 10:28 pm
I wrote a simple multi-threaded C version of the pentanomial simulator.

https://github.com/vdbergh/simul

Everything in a single C file. As it is much much faster than the Python version one can see better how accurate the implementation is.
Now with a decent README.md!
by Michel
Sun Feb 02, 2020 10:28 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Stockfish Reverts 5 Recent Patches
Replies: 9
Views: 2846

Re: Stockfish Reverts 5 Recent Patches

I wrote a simple multi-threaded C version of the pentanomial simulator.

https://github.com/vdbergh/simul

Everything in a single C file. As it is much much faster than the Python version one can see better how accurate the implementation is.
by Michel
Sun Feb 02, 2020 9:53 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Stockfish Reverts 5 Recent Patches
Replies: 9
Views: 2846

Re: Stockfish Reverts 5 Recent Patches

The validity of the pentanomial model can be verified by simulation. https://github.com/vdbergh/pentanomial Concerning short tests: there are various things to consider, notably: Fishtest Elo bounds are no longer BayesElo. The stopping time distribution for an SPRT has long tails. The great majorit...
by Michel
Thu Jan 23, 2020 7:13 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Hyatt 4000 openings
Replies: 7
Views: 1702

Re: Hyatt 4000 openings

Unbalanced opening are completely safe to use as long as the pentanomial variance is used to calculate the error bars. For SPRT it is bit more complicated but pentanomial SPRT has been integrated into Fishtest recently and is working satisfactorily. So now one can just copy that code (although it is...