Software for comparing final game position with result?

[John Saunders]

The query about the result of the Sivanandan-Wells game (discussed here - now resolved), as to whether it was a resignation or loss on time prompted the thought that we need a quick way to test the final positions of game scores to see if they match the recorded result of the game. Does such software exist, either as part of database software (ChessBase, Hiarcs, SCID, etc) or other chess utilities such as ChessSuite?

What I have in mind is something that would sweep through a database and flag those games with a final position where the recorded result does not tally with (say) a Stockfish assessment. Does such a facility already exist? I'm aware of software with blundercheck facilities but these apply to each move of a game and I'm only interested in the final position.

Such a facility would be very handy for those of us who curate databases. It would allow us to focus on each flagged game in turn and investigate further, to see if the game ended in (a) a time or mobile phone loss; (b) a wrongly-recorded result; (c) a faulty or incomplete score; (d) a misjudged resignation/draw decision by a player; (e) some other incident.

[Ian Thompson]

I asked about a specific example of this sort of thing - spurious king moves to the centre of the board at the end of the game - a few years ago but no-one had an answer.

[Roger de Coverly]

- spurious king moves to the centre of the board at the end of the game

I think this arose in one of my games in the British Over 65 last week. If one player resigns then the clock is usually stopped which tells the DGT board to cease or pause recording. If the game ends with "draw agreed", then it's possible that the clocks aren't immediately stopped. When a King is moved to the centre of the board to indicate a result, then it might be a legal move and thus gets recorded.

[John Saunders]

Spurious moves at the end aren't so much of a problem for the database curator as the DGT software usually outputs them as text rather than legitimate moves, which means they are easily identified as games with annotations when you perform a Chessbase (or similar) list of the database. I did this with the British Championship games after downloading them from chess-results.com, went through the various games flagged with annotations (there were a dozen or so of them) and in all cases (I think) simply deleted the spurious text found at game-end.

Please let's not get into a debate about whether DGT should add a result button to their clocks, but try and stick to the point. (yes, I know, a triumph of hope over experience!)

[Ian Thompson]

Spurious moves at the end aren't so much of a problem for the database curator as the DGT software usually outputs them as text rather than legitimate moves, which means they are easily identified as games with annotations when you perform a Chessbase (or similar) list of the database. I did this with the British Championship games after downloading them from chess-results.com, went through the various games flagged with annotations (there were a dozen or so of them) and in all cases (I think) simply deleted the spurious text found at game-end.

Yes, if the move would have been illegal and consequently gets recorded as text it's easy to find those games without much effort.

Do you think a legal king move to the centre of the board really was the last move in any of these games?

• Adams,H-Pert,M, Championship, Rd. 6
• Liu-Berkley, Major Open, Rd. 2
• Peart-Shankar, Major Open, Rd. 6
• Rushbrooke-Wall, Major Open, Rd 6
• Armstrong-Jerome, Over 65, Rd. 5

[John Saunders]

Do you think a legal king move to the centre of the board really was the last move in any of these games?

• Adams,H-Pert,M, Championship, Rd. 6
• Liu-Berkley, Major Open, Rd. 2
• Peart-Shankar, Major Open, Rd. 6
• Rushbrooke-Wall, Major Open, Rd 6
• Armstrong-Jerome, Over 65, Rd. 5

Yes, I'm aware it becomes a problem when the spurious move at the end happens to be legal. It is another example of why a software utility to detect questionable game-ends is highly desirable. Presumably you have discovered these examples by going through the games manually - or perhaps searching for Ke4/Ke5/Kd4/Kd5 last moves?

This particular problem is one where the database curator's judgment (that an improbable Ke4/Ke5/Kd4/Kd5 move is spurious) can probably be relied upon to be right and the move simply deleted when found. It's the finding that is the problem.

[Ian Thompson]

Presumably you have discovered these examples by going through the games manually - or perhaps searching for Ke4/Ke5/Kd4/Kd5 last moves?

I'm not aware of any way of searching for any particular move being the last move of the game. I searched for games where the king went to one of those squares during the game and then manually looked for games where it was the last move of the game. It's not practical to do that on large numbers of games. It's not a foolproof way of finding all such games. I might have missed games where the king went to the centre of the board earlier in the game and then back to the centre at the end of the game (because ChessBase jumps to the first position in the game meeting the search criteria). I was probably wrong to include Rushbrooke-Wall in the list of games because 43...Kd5 is a reasonable move to play in the position, although many other moves also win for Black.

[Alex McFarlane]

You could try teaching players that if the king move to indicate the result is legal then they should simply remove a piece before positioning the kings.

Good luck with that one, though

[Christopher Kreuzer]

Do you have to place the kings simultaneously? Can you not move the 'wrong' king (i.e. the king of the player who does not have the move) to the appropriate square first, followed by the other one? I suppose there is also an issue if one of the kings is already on the square it needs to move to, though maybe not, as you just move the other king to the adjacent square (which will automatically be an illegal move).

(I don't actually know the convention anyway, so I don't know why I am commenting! Something about same colour squares = draw, if kings on the central white squares that is a White win, if on the central black squares that is a Black win? Am guessing here...)

[Christopher Kreuzer]

What I have in mind is something that would sweep through a database and flag those games with a final position where the recorded result does not tally with (say) a Stockfish assessment. Does such a facility already exist? I'm aware of software with blundercheck facilities but these apply to each move of a game and I'm only interested in the final position.

You would have to decide how widely to set the drawing parameters... It would be a useful tool. I am almost certain some databases will already have "final position" assessments in them. That bit is trivial. Many people agree draws in +1.5 or -1.5 positions though (or even greater values). That would mean you would get a lot of edge cases.

[GrahamStuart]

The query about the result of the Sivanandan-Wells game (discussed here - now resolved), as to whether it was a resignation or loss on time prompted the thought that we need a quick way to test the final positions of game scores to see if they match the recorded result of the game. Does such software exist, either as part of database software (ChessBase, Hiarcs, SCID, etc) or other chess utilities such as ChessSuite?

What I have in mind is something that would sweep through a database and flag those games with a final position where the recorded result does not tally with (say) a Stockfish assessment. Does such a facility already exist? I'm aware of software with blundercheck facilities but these apply to each move of a game and I'm only interested in the final position.

Such a facility would be very handy for those of us who curate databases. It would allow us to focus on each flagged game in turn and investigate further, to see if the game ended in (a) a time or mobile phone loss; (b) a wrongly-recorded result; (c) a faulty or incomplete score; (d) a misjudged resignation/draw decision by a player; (e) some other incident.

The games exported from Lichess seem to have the final evaluation. I copied these into Excel and wrote a quick macro to filter on the game. Additional filters for 1-0, 1/2 and 0-1 added and the eval checked (manually but this could be programmed). The manual check does not take long as the games can then be checked to highlight potential issues. The main ones which came out were probably draws where one player was better, but this would highlight games with low evaluation with a positive result. Although in the championship I did not see any.

This was from this game where drawn but plus 2.7
https://lichess.org/broadcast/2025-brit ... w/gvqB7OII

If you have any particular results which you think are strange I can check from the British or other events. I was surprised the Lichess had the evaluation included. This is under the Download pgn on the page.
https://lichess.org/broadcast/2025-brit ... 1/Rus7vgww

Not sure if this is that useful, but may help in someway?

[Geoff Chandler]

What I have in mind is something that would sweep through a database and flag those games with a final position where the recorded result does not tally with (say) a Stockfish assessment.

Hi John,

I know the problem exists and spotted a few in my time. I was wondering how common it was and can we not just let it go as 'just one of those things and ignore it' (which BTW is the Chandler Motto)
So I had a look at all the games from chessgames.com, for the year 2024 that were 1-0, the move Kd5 had been played and the last move was Kd5 (or Kxd5)
Over 700 popped up and I found 52 what I would confirm as bogus KD5 resignations - And that was looking for 1-0 from one year. Half of dozen or so that were probably also a bogus KD5 resignation but there was a slight doubt. Black played Kd5 realised playing on was pointless so resigned. (they were probably all bogus.)

TBH I was looking for any reason not to classify it as a bogus KD5 resignation, because I was getting fed up logging them (it's a Chandler Motto thing)
I am not going to send 50-60 correction slips to chessgames.com (once the Chandler Motto takes a hold you never shake it off) If I do I'll have to do the same for every year since 2000 for both colours. Sod that.

I'll send in one for a Samuel Sevian game because it features a Kxd5.

https://www.chessgames.com/perl/chessgame?gid=2812111

8/5R2/3k2K1/3P2P1/8/8/8/5r2 b - - 2 76

Black played 76...Kxd5 and resigned.

[Matt Mackenzie]

That's a mildly interesting one - it is likely another superfluous move but could also, just possibly, be an oversight.

(which would fit with resigning before the opponent can reply, so that doesn't prove anything either)

[Phil Makepeace]

I've had one which has been annoying me for 21 years.

I played a horrific game but got away with it, and my opponent offered a draw in the final position as his extra piece is falling. On the database it's recorded as 1-0 but it was at least graded correctly.

[NickFaulks]

Possibly unrelated to the current discussion, but do you think this result is correct?

https://lichess.org/broadcast/fujairah- ... 6/zKvBbyW6

It looks like a perfectly good place to offer a draw, even if the computer does like White.