Re: TCEC Division 1 results simulator

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

Uri Blass
Posts: 10282
Joined: Thu Mar 09, 2006 12:37 am
Location: Tel-Aviv Israel

Re: TCEC Division 1 results simulator

Post by Uri Blass »

Laskos wrote: Mon Sep 03, 2018 12:01 pm
chrisw wrote: Mon Sep 03, 2018 11:54 am
Laskos wrote: Mon Sep 03, 2018 10:44 am
chrisw wrote: Sun Sep 02, 2018 11:30 pm
Laskos wrote: Sun Sep 02, 2018 10:59 pm
chrisw wrote: Sun Sep 02, 2018 7:06 pm
Laskos wrote: Sun Sep 02, 2018 5:17 pm
chrisw wrote: Sun Sep 02, 2018 4:30 pm

Code: Select all

Division Premier sim predictions, percent chances of engine finishing in Nth place
Engine     Tournament Initial  First   Second  Third   Fourth  Fifth   Sixth   Seventh Eighth
Stockfish      3511   3511     0.700   0.220   0.064   0.013   0.003   0.001   0.000   0.000
Houdini        3470   3470     0.215   0.447   0.225   0.079   0.025   0.006   0.001   0.000
Komodo         3440   3440     0.069   0.232   0.380   0.194   0.087   0.029   0.007   0.002
Ethereal       3401   3401     0.011   0.061   0.176   0.317   0.242   0.127   0.051   0.016
Fire           3386   3386     0.005   0.032   0.110   0.244   0.297   0.191   0.088   0.033
Chiron         3356   3356     0.001   0.007   0.034   0.105   0.207   0.313   0.215   0.118
Ginkgo         3322   3322     0.000   0.001   0.006   0.028   0.079   0.182   0.331   0.374
Andscacs       3315   3315     0.000   0.000   0.004   0.019   0.061   0.151   0.307   0.457
Again Ethereal too high in the last table. Probably Chiron too.
This could well be. They both got a lot of tournament elo from being successful in Division 1. This was added in to their start elos at a 50% weighting. I guess it remains to be seen if this was too much (you can claim told you so rights). Or maybe Ethereal or Fire can steal third place, the sim gives about 0.25 for either of then to get third.
So, all places combined, you seem to get around 50% that top 3 will not be the "big three" in the Premiere Division (56 games each). Want a fair 1:1 bet against my statement that they will be?
is about 33% that Ethereal and 14% for Fire to get 3rd place or better. According to all the assumptions, elo, spin of the dice bla bla. Gimme 3 to 1 and you're on.
I don't understand, you have 47% (according to you) and want 3:1? That 33% for Ethereal in top 3 is ridiculous. Instead of a fair 2:1, I can go even for 5:1 that this won't happen. Deal?
I have Ethereal 1% chance to be first, 6% second and 17% third = 24% to be in top three. Disregard my earlier 33% for Ethereal, my bad adding up or something.

So, that's 35% for either Fire or Ethereal in top three. Okay I take 5:1 on an upset in the top three.
No, 5:1 for Ethereal in top 3, that was my proposal, maybe I haven't expressed it very clearly. Your prediction is 24%, or close to 3:1, so 5:1 is a good proposal from me. I can go even to 7:1 (Ethereal in top 3). Deal?
With Fire engine I have some issues, as it can be improved fast, Kranium style.
I do not know why you assume Fire can be improved fast and Ethereal but here is what I read in the TCEC chat

"The updated engines for Season 13 Premier Division are Stockfish, Komodo, Andscacs, Ginkgo and Ethereal. Confirmed same versions: Houdini and Fire."
User avatar
Laskos
Posts: 10948
Joined: Wed Jul 26, 2006 10:21 pm
Full name: Kai Laskos

Re: TCEC Division 1 results simulator

Post by Laskos »

Uri Blass wrote: Mon Sep 03, 2018 12:30 pm

I do not know why you assume Fire can be improved fast and Ethereal but here is what I read in the TCEC chat

"The updated engines for Season 13 Premier Division are Stockfish, Komodo, Andscacs, Ginkgo and Ethereal. Confirmed same versions: Houdini and Fire."
Good, so I can make some bets on Fire too.