Chess Masters: the Endgame

[Christopher Kreuzer]

It might be possible to put those starting positions (even though some are illegal) into a convenient online chess engine to see who is theoretically better. If this is not possible, does anyone know of a chess engine that will allow these positions and analyse them?

[LawrenceCooper]

I think to have a lot of knights is good as at that level with a fast time limit the opponent will fall for a bunch of knight forks.
The Thalia v Lula match up tested that out as Thalia chose 3 bishops and won, but theoretically I’m not convinced that having multiple same coloured bishops is good.

It would certainly make checkmating with two bishops somewhat problematic

[Christopher Kreuzer]

I was wondering what potential there is for fortresses when you have 7 knights on the board? But having eight pawns each as well makes it a bit tricky. Am about to see which of the legal positions are winning or not... (oh, and the illegal positions can be set up in Lichess!).

[Christopher Kreuzer]

2. Caitlin vs Richie

Board 2: 3-man vs 5-man army (both 23 points)
Richie: 2rqkq2 (a8,b8,g8,h8=empty squares)
Caitlin: B1QRKB1B (b2,g1=empty squares)

2rqkq2/pppppppp/8/8/8/8/PPPPPPPP/B1QRKB1B

Apparently, the assessment is +1.7 for White, with a slight advantage for White unless the f-pawn is moved when Black has the advantage instead. If Black moves first, the engine still thinks White has an advantage of about a pawn (this may not be precisely the same as swapping colours).

Obviously in practice, anything can happen.

[Christopher Kreuzer]

1. Thalia vs Lula

Board 1: 6-man vs 6-man army (both 24 points)
Lula: bnnqkbn1 (h8=empty square)
Thalia: BNBQKBN1 (h1=empty square)

bnnqkbn1/pppppppp/8/8/8/8/PPPPPPPP/BNBQKBN1

Unsurprisingly, the assessment is about 0.5 for White regardless of who plays first. Slightly higher assessment for White if White plays first.

[Christopher Kreuzer]

3. Navi vs Kel

Board 3: 4-man vs 4-man army (both 24 points)
Kel: 1nq1kqn1 (a8,d8,h8=empty squares)
Navi: 1N1QKQN1 (a1,c1,h1=empty squares)

1nq1kqn1/pppppppp/8/8/8/8/PPPPPPPP/1N1QKQN1

Very small edge (i.e. practically non-existent) for White. Slight change towards equality if Black to move. Maybe it is better to start with two queens on the same square?

[Christopher Kreuzer]

For the record:
nnnnknnn/pppppppp/8/8/8/8/PPPPPPPP/BBBBKBBB w - - 0 1

With White to move, the 7 bishops are +4.5, while if it is Black to move, the 7 knights are still losing, but only by +2.4.

And, somewhat astonishingly, for 7 knights against 2Q and R, the assessment is almost equal:

2rqkq2/pppppppp/8/8/8/8/PPPPPPPP/NNNNKNNN w - - 0 1

That is only -0.3 with White to move. With Black to move, it is -1.3. Shove the heavy black pieces to the side of the board, and in at least one position White starts to be ahead.

Of course, all these assessments would need to be done properly with advanced engines and the right settings, running for more than just a few seconds (I used the online Lichess Stockfish 16 at 7MB NNUE and only for a few seconds - only have the faintest idea what that means).

[Christopher Kreuzer]

This was my incredibly pathetic attempt to win with 7 knights. To be fair, I should have selected a lower SF level...
[Event "Casual correspondence game"] [Site "https://lichess.org/hhXVumpd"] [Date "2025.04.22"] [White "Anonymous"] [Black "lichess AI level 8"] [Result "0-1"] [GameId "hhXVumpd"] [UTCDate "2025.04.22"] [UTCTime "20:36:17"] [WhiteElo "?"] [BlackElo "?"] [Variant "From Position"] [TimeControl "-"] [ECO "?"] [Opening "?"] [Termination "Normal"] [FEN "rq2k2q/pppppppp/8/8/8/8/PPPPPPPP/NNNNKNNN w - - 0 1"] [SetUp "1"] [Annotator "lichess.org"] 1. Nhg3 h5 2. h4 g6 3. Nh3 d5 4. f4 e5 5. fxe5 Qxe5 6. Nd3 Qd4 7. c3 Qxh4 8. N1f2 b6 9. Nc2 c5 10. b3 Qhd8 11. Nba3 h4 12. Ngh1 Qb7 13. Nce3 Kf8 14. Nac2 d4 15. Neg4 Rc8 16. Ndf4 Qa6 17. a4 b5 18. Ne5 bxa4 19. bxa4 g5 20. Nh5 Qxa4 21. Nfg4 Qxc2 22. Nhf6 dxc3 23. Nfd7+ Kg7 24. Ngf6 Qa5 25. Nh5+ Kh6 26. Nxf7+ Kxh5 27. g4+ Kg6 28. Nfe5+ Kg7 29. Nxg5 cxd2+ 30. Kf2 d1=Q 31. Ne6+ Kh8 32. Nf7+ Kg8 33. Nh6+ Kh7 34. g5 Qcxe2+ 35. Kg1 Qdxf1# { Black wins by checkmate. } 0-1

Analysing the mistakes there, it is quite instructive to watch the engine show you how to play with an army of 7 knights. Not that it will ever be of use in a practical game!

[Christopher Kreuzer]

I just watched the full Eliminator game for Episode 7, and there are some funny one-liners, but the chess wasn't great (which meant there was not really much for the commentators to talk about). There was the potential for a game, but the mind games took over and the game was pretty much over as a contest after the first mistake.

I hope anyone who has not watched Episode 7 yet will not be spoilered by the poll I am about to start on who might win next Monday (Episode Eight).

[Mark Hannon]

I would bet on Thalia but it’s pretty open. Kel gets a lot of camera time but he has made mistakes though his antics may put off the other players.Richie may be the solid one , Lula seems too weak in the opening to win the whole thing . I wonder if it will be classical chess and semis and a final for the last 2 games or more puzzles and variants

[Christopher Kreuzer]

There are five puzzles seen in the preview of the final at the end of Episode 7. I doubt the camera will pause long enough in the final for viewers to solve them though... You see some of the players attempting the puzzles, but not all. Maybe these puzzles follow an initial elimination round?

[Mark Hannon]

There are five puzzles seen in the preview of the final at the end of Episode 7. I doubt the camera will pause long enough I'm the final for viewers to solve them though... You see some of the players attempting the puzzles, but not all. Maybe these puzzles follow an initial elimination round?

Surely the very last event will be a 1 to 1 normal chess game? Maybe they’ll eliminate 2 players via puzzles then have a final game? Or the puzzles will be the prelude to another 2 vs 2 team game then a final?

[Christopher Kreuzer]

Agree the series needs to end with a 1-to-1 normal chess game. Hopefully not a damp squib!

Only five votes so far in the poll.

Vote over here! viewtopic.php?f=2&t=14661

[Roger Lancaster]

Only five votes so far in the poll.

Vote over here! viewtopic.php?f=2&t=14661

Hmm. Is there a "Who cares?" voting option?

[Christopher Kreuzer]

Only five votes so far in the poll.

Vote over here! viewtopic.php?f=2&t=14661

Hmm. Is there a "Who cares?" voting option?

I thought that might be a bit cruel. I did also consider phrasing it as who do you WANT to win?

I see Jon Mahoney has commented in that thread (converted should be coveted). It is true that not many people have bothered to vote, so maybe that is your answer.

In other news, Leonard Barden has covered the decline in viewing figures in his column (he says maybe due to Easter holidays).

https://www.theguardian.com/sport/2025/ ... style-open

Unfortunately, Kel has been misnamed as Kai (there was an earlier contestant named Cai).

To be fair, Roger and Jon have a point. Strictly levels of coverage this show does not have! But then that was never the expectation.

It will be interesting to see the viewing figures for the final.