Free online Chess server. Play Chess now in a clean interface. No registration, no ads, no plugin required. Play Chess with the computer, friends or random opponents.
Sign in
Reconnecting
  1.  Improving Horde
  2. Forum
  3. Stockfish's "opinion" on balancing horde chess

@ubdip I see.

OK, so I guess the match proved quite conclusively that strong human players are just going to crush with this White position. How about the next match...maybe try the centered setup? Or the regular Horde setup + d5 OR e5

I do not like the setup with one additional pawn, mainly because of the asymmetry but also because it still seems to favor white too much. I would go for a setup with advanced central or rook pawns. For the further we already have the data (~60%), and the latter might be even more balanced.

@ubdip Great job in providing those data !

I tend to like the advanced setup, as I think giving one extra pawn or more to White gives too much of an advantage to White. Also, there could be different versions of the advanced setup: leaving two unoccupied central squares between the 1st and the 4th rank.

But of course the version you proposed of the advanced setup should be the first one to be tried out.

@svenos
I also like the advanced setup. Advancing the rook pawns instead of the central pawn should be a slightly smaller advantage and preliminary results seem to support this. The final results should be available in a few hours (see http://35.161.250.236:6543/tests ), I'll again summarize them here.

The results of the advanced rook pawns setup are in:

advanced rook pawns
rnbqkbnr/pppppppp/8/PPP2PPP/PPPPPPPP/PPPPPPPP/PPPPPPPP/1PPPPPP1 w kq - 0 1
----------------------------------
10"+0.1"
62.2%

30"+0.3"
57.4%

90"+0.9"
62.4%

So I think both setups with advanced pawns (central or rook pawns) are good candidates as the score is quite stable around 60% for both of them.

I'll now investigate the change in the score when we flip colors.

Flipping the colors might be confusing at first, but it's the way to go trying to approach a kind of equality.

One question anyhow still remains open that can not be answered by any of these tests: Is Stockfish's playing strength high enough to make these results meaningful?

There is not that much data, but it looks to me like Stockfish level 8 is at a level of about 2200-2300 when the human opponent uses a blitz time control. The strength of level 8 should be somewhere between 10''+0.1'' and 30''+0.3'' time controls. From that I assume that at 90''+0.9'' it should be at the same level as the top players, but since I am only deducing this, the estimate might be inaccurrate.

By testing at different time controls I try to mitigate this uncertainty, since the change in the score should usually give hints what the score might be at longer time controls (i.e., higher level).

www.youtube.com/watch?v=nH-DA321wSU

Hmm yes SF does seem to play at quite a low level. It can't beat svenos in a match, right?

@FischyVishy
Stockfish level 8 would probably lose a match against svenos. With equal time controls for human and engine on an average PC Stockfish should be competitive.

Thx alot FischyVishy for the vid, great stuff!
really intresting to hear some thoughts of other people! :)

This topic has been archived and can no longer be replied to.