CDrill 1800

[Ferdy]

Site:
https://sites.google.com/view/cdrill

Download:
https://sites.google.com/view/cdrill/download

[PK]

nice project! congratulations!

[Ferdy]

nice project! congratulations!

Thanks, you can post here your engine's setting to simulate a human play at 1800 elo.

[MikeB]

nice project! congratulations!

Thanks, you can post here your engine's setting to simulate a human play at 1800 elo.

For stockfish, limiting nodes to about 4K , will give provide roughly 1800 ELO, more nodes ( 2x or more ) for most other engines.

[Ferdy]

nice project! congratulations!

Thanks, you can post here your engine's setting to simulate a human play at 1800 elo.

For stockfish, limiting nodes to about 4K , will give provide roughly 1800 ELO, more nodes ( 2x or more ) for most other engines.

Thanks I will try that.

[MikeB]

nice project! congratulations!

Thanks, you can post here your engine's setting to simulate a human play at 1800 elo.

For stockfish, limiting nodes to about 4K , will give provide roughly 1800 ELO, more nodes ( 2x or more ) for most other engines.

Thanks I will try that.

Just to clarify for those in the US , this is not USCF 1800, but more like FIDE 1800 - on a USCF basis it might be closer to 2000.

[Ferdy]

nice project! congratulations!

Thanks, you can post here your engine's setting to simulate a human play at 1800 elo.

For stockfish, limiting nodes to about 4K , will give provide roughly 1800 ELO, more nodes ( 2x or more ) for most other engines.

I test CDrill 1800 (around 2000 FIDE Elo) vs Sf 8 Ladder Rating 2000 and this is the result in blitz.

Code: Select all

Score of CDrill 1800 vs Stockfish 8 Ladder 2000: 2 - 44 - 4 [0.080]
Elo difference: -424.28 +/- 203.45

50 of 50 games finished.

The gap is too big for this 50 games.

CDrill 1800 is around FIDE 2000 after couple of estimate and test vs Mephisto Amsterdam.
http://www.talkchess.com/forum/viewtopic.php?t=62643

[PK]

Rodent should be close to CDrill at about 2000 nodes per second. The problem is that in my book it is also 2000 Elo.

I watched a couple of games, and CDrill's midgame performance looks as it should, however there is a big problem with early queen moves in the opening. Since this is the engine to play against humans, I'd propose to add the following heuristic: if the queen is not on its starting square, apply a penalty of, say, 3 centipawns for each minor piece on its starting square and for d/e pawns on their starting squares.

@MikeB, I've got a question: at what time control does Your Stockfish perform at 2000 Elo? Your approach has a virtue of being a perfect anchor point, but it does not take into account the fact that playing strength increases with time control. Having said that, nps reduction suffers from the inverse effect: at longer time limits a human player suffers no matter if engine runs at 100.000 or 200.000 nps.

[Ferdy]

Rodent should be close to CDrill at about 2000 nodes per second. The problem is that in my book it is also 2000 Elo.

I watched a couple of games, and CDrill's midgame performance looks as it should, however there is a big problem with early queen moves in the opening. Since this is the engine to play against humans, I'd propose to add the following heuristic: if the queen is not on its starting square, apply a penalty of, say, 3 centipawns for each minor piece on its starting square and for d/e pawns on their starting squares.

Good point, I will look into it. I also plan to add book support even for up to 4 or 5 moves only.

[mar]

Very nice project Ferdy!

Care to explain a bit more about how you try to make CDrill play more human-like? I'm allergic to this kind of mantra but I know if you try to make something play human-like it's way more
than random tweaking of piece values.

I often play my old iOS chess app built on top of cheng3 and I have to say I messed up with difficulty scaling.
Adding noise to eval without tuning it is a disaster, I can win ~30% against "skill 99" but I only have 2 draws out of about 100 against skill 100 (full power) - my bad,
the good thing is it kills my ego

So I still think that maybe time odds scaling as I do in cheng4 is the way to go (without expensive tuning).

Personalities sound really good but there's still the tweaking issue to match real strength.

That being said, I'm a super-lousy OTB player, but it's fun to play engines that match your (=mine) level.

I also wanted to ask - do you have some pseudo-random book with a lot of variation?
I'm tempted to start a mini-project with big chessboard and one engine plus high-variety book,
but as usual motivation is #1 problem, though I'm tired of elo-hunting I admit