Talk:Pawnless chess endgame

Absurdity
[Queen and a rook versus a queen and a rook]: "Despite the equality of material, the player to move wins in 67.74 percent of positions". This cannot be - the 2:1 advantage cannot switch from one side to the other at each move. AMackenzie (talk) 11:13, 30 June 2008 (UTC)


 * The one that moves first has the advantage. He can usually keep up an attack.  Bubba73 (talk), 15:34, 30 June 2008 (UTC)


 * See the footnote I just added. Bubba73 (talk), 15:58, 30 June 2008 (UTC)


 * It is, in fact, not absurd at all. Let's say White is to move; in a randomly generated position there is a 2/3 chance White can win. After White's move, the position is not randomly generated and so the previous statement does not imply there is a 2/3 chance Black can win. White aims to leave, after each move, a position belonging to the minority that is still winning for White. Since the queen and rook have great mobility on an open board, chances are in his favour that he will be able to do this. 91.105.33.98 (talk) 15:35, 19 December 2010 (UTC)


 * That is a good explanation, thanks. (And if White was not winning, he certainly would want to avoid moving into a position where he would be losing.) Bubba73 You talkin' to me? 15:42, 19 December 2010 (UTC)

Botvinnik-Tal, 1961
Tal talked about this position in his book The Life and Games of Mikhail Tal, but I can't find it again. The score of the game was not given, nor the position. But he talked about going into it because he had encountered this before (on the short end, I believe). If what he says about it can be located, it might make a good addition to the article. Bubba73 (talk), 02:46, 2 July 2008 (UTC)

Summary Table
According to the summary table, every pawnless endgame listed except one is described as "easy" despite the fact that some are described as being difficult or tricky in this article's notes (19,20). Furthermore, the Wikipedia article on the Cochrane Defense describes it as difficult, and it is easy according to the table. Also, many of the endgames listed on the table do not have notes referenced. — Preceding unsigned comment added by 74.243.102.152 (talk • contribs)


 * You are right. All of those assessments in the table come from one source.  There are some footnotes from more expert players (e.g. Nunn) that say differently.  It may be best to take those assessments out.  Bubba73 You talkin' to me? 16:06, 1 February 2012 (UTC)

Rook versus rook?
Rook versus rook is not mentioned. Shouldn't it be? Magic9mushroom (talk) 14:27, 28 May 2012 (UTC)


 * It isn't in any of my three main references: Secrets of Pawnless Endings, Fundamental Chess Endings, and A Pocket Guide to Chess Endings. But it will be a draw except when there is a quick checkmate or win of the other rook.  Bubba73 You talkin' to me? 17:12, 28 May 2012 (UTC)


 * I figured as much, but it should still be mentioned given that two knights vs. king is. Magic9mushroom (talk) 17:21, 28 May 2012 (UTC)


 * There is a lot more to say about Two knights endgame, and it is referenced. So far I haven't found a reference for rook versus rook. It is sort of surprising that it isn't in those references but queen versus queen is, but Q vs Q does have deeper and more subtle lines. Bubba73 You talkin' to me? 19:09, 28 May 2012 (UTC)


 * OK, I found a reference and added it under "major pieces only". Bubba73 You talkin' to me? 20:02, 28 May 2012 (UTC)

Why two bishops vs. a lone king win only in 99,97 per cent cases?
I have already understood that bishop + knight win in 99,5% (not 100) because the defending king can give a fork on them if they are poorly placed at the beginning of the endgame. But what about two bishops endgame? Can someone explain why 99,97% but not 100? — Preceding unsigned comment added by Alex Chorny (talk • contribs) 14:07, 14 October 2012 (UTC)


 * That's a good question. If White is to move, the two bishops are on opposite color, and it is a legal position, it can't be a fork. I can't think of a position that is not a win for the bishops.  I'll ask around.  Bubba73 You talkin' to me? 15:37, 14 October 2012 (UTC)

Oh, forgive me for my commas. Of course, there should be dots in my question: 99.97 instead of 99,97. (I made this mistake because our country uses comma as decimal separator.) — Preceding unsigned comment added by Alex Chorny (talk • contribs) 19:58, 14 October 2012 (UTC)

Also a draw with the bishop on c4, d5, e6, f7, or g8 instead of b3. Also reflected with respect to the diagonal, and in the other corners. Bubba73 You talkin' to me? 22:07, 14 October 2012 (UTC)


 * In addition, the white king can be on d2 or d1. Bubba73 You talkin' to me? 22:17, 14 October 2012 (UTC)

Queen versus rook
The section queen versus rook is too short and doesn't present enough interesting features. — Preceding unsigned comment added by Nzarathoustra (talk • contribs) 11:38, 20 November 2012 (UTC)


 * I don't think it is too short, but rather quite long with many examples that are quite confusing for the novice, who might want to find a simple description of the "winning process" instead of lots of "random information" which is of not much use without knowing what would be the correct play. &mdash; MFH:Talk 02:27, 5 February 2022 (UTC)

Rook and Knight versus Bishop
Is missing - FWIW, intuitively, I would say this is a win for Rook + Knight, but I couldn't find a definitive answer anywhere (178.12.33.158 (talk) 14:37, 2 October 2016 (UTC))


 * I'm pretty sure that it isn't in the references that were used. However, the rook and knight should be able to at least force the exchange of the minor pieces and then win.  Bubba73 You talkin' to me? 15:07, 2 October 2016 (UTC)

7-men tables
This table shows seven-piece endgames.

I have 1-year subscription of Lomonosov Tablebases and i can add the remaining info. Sunny3113 (talk) 18:37, 27 March 2016 (UTC)


 * I like what you did. ClueBot reverted it, but I reported that as a false positive.  Bubba73 You talkin' to me? 22:22, 27 March 2016 (UTC)


 * Thank you, i will add more combinations. Sunny3113 (talk) 12:59, 28 March 2016 (UTC)


 * Yes, it is interesting. I haven't seen these seven-piece results before.  Bubba73 You talkin' to me? 17:28, 28 March 2016 (UTC)

Are some of these reversed as to who wins? For instance Q+N versus three rooks. It seems like the winner should be the three rooks. There are some others too. Bubba73 You talkin' to me? 17:49, 28 March 2016 (UTC)

In many cases one of the defendant pieces get captured straight away in just few moves. For example here FEN: 5q2/R6n/8/8/6k1/8/3K3R/R7 b - - | ...Qd6+ and Qxh2 next with easy six-piece win. Sunny3113 (talk) 21:54, 28 March 2016 (UTC)
 * I understand this would probably take more work, but surely it should be possible to exclude such cases by cutting out those with very early conversions? Positions where conversion happens quickly due to a piece being tactically lost are surely not relevant to the average evaluation of the 7-man endgame in question. Double sharp (talk) 14:31, 13 April 2019 (UTC)

Three minor pieces versus two minor pieces?
This is not included in the list. I imagine it should be a draw in most cases, but with 3 minor pieces versus 1 included it seems to be a natural follow-up question. Double sharp (talk) 15:36, 16 September 2017 (UTC)


 * It wasn't in any of the sources I used, but I'm pretty sure that generally it would be a draw. If a pair of the poeces is exchanged, it is down to 2 minor pieces versus 1, which is normally a draw except for two bishops vs one knight.  Bubba73 You talkin' to me? 16:51, 16 September 2017 (UTC)
 * I haven't found any good sources for this either. OTOH, after a few searches I found that Marc Bourzutschky (one of the "wizards of 7-men endgames", as Tim Krabbé puts it) once addressed this in a post on the Rybka Forum, in which he claimed that the only generally winning cases of this endgame are BBN vs NN and BBN vs BN (assuming the bishops are on opposite colours). I am not sure if this is reliable enough, though, despite the authority of the poster; it may yet fall under the third exception under WP:USERGENERATED. What do you think? Double sharp (talk) 13:06, 17 September 2017 (UTC)
 * Well, I have added it with this source, even if not ideal, because of the authority of the poster and the interestingness of the result. Double sharp (talk) 01:13, 25 February 2019 (UTC)


 * Thank you for adding that - I think it is good. Bubba73 You talkin' to me? 01:44, 25 February 2019 (UTC)

FWIW, this helpful game collection of chessgames.com includes two games which ended in BBN vs BN (both wins for the BBN side according to Lomonosov Tablebases). The first was a draw; the second was won by the stronger side exchanging his N for the opposing B, leaving a won BB vs N endgame. Double sharp (talk) 14:29, 13 April 2019 (UTC)

The 3 rooks vs 2 rooks incident
I removed this due to lack of sourcing, but Paul Lamford has confirmed that the incident did indeed happen. Possible sources include Tim Krabbe's Chess Curiosities and Europe Echecs. MaxBrowne2 (talk) 03:06, 3 February 2019 (UTC)


 * I didn't find it at Krabbe's site. And it isn't in chessgames.com. Bubba73 You talkin' to me? 03:41, 3 February 2019 (UTC)


 * This was the event, but the game does not appear in the databases. MaxBrowne2 (talk) 04:06, 3 February 2019 (UTC)


 * I know that we are supposed to use secondary sources, not primary sources. But in this case, I'll accept a primary source, i.e. a personal communication from Lamford.  It would be nice if he has a scoresheet or PGN file.  Bubba73 You talkin' to me? 05:15, 3 February 2019 (UTC)
 * No we can't do that. It needs to be published somewhere. MaxBrowne2 (talk) 05:31, 3 February 2019 (UTC)


 * The editor that made that edit is anonymous, so they knew about it somehow, but they probably aren't around to ask. Bubba73 You talkin' to me? 06:18, 3 February 2019 (UTC)

Two rooks versus four minor pieces?
I realise that we are unlikely to have any information about 8-man endgames like this one, but since Q ~ 2R and Q vs 4 minors is included, it seems a natural enough question. (I would guess that it is generally a draw if they are the normal four minor pieces as the stronger side can sacrifice his rooks for the opponent's bishops.) Double sharp (talk) 02:05, 27 February 2019 (UTC)


 * I think that is probably true. The two rooks have more defenses against four minor pieces than a queen does.  Bubba73 You talkin' to me? 02:21, 27 February 2019 (UTC)

Frequencies of such endgames?
It would be interesting to know how commonly these endgames appear. I would guess from personal experience and observing the results of computer self-play that anything other than the basic checkmates, Q vs R, R+minor vs R, and R vs minor is actually very rare (I got through thousands of computer self-play games and I only found a 7-man pawnless endgame once); I wonder if there are any actual statistics given in reliable sources. Double sharp (talk) 02:11, 27 February 2019 (UTC)


 * I don't think I ever had any except for ones that are basic checkmates. But my daughter had some in tournaments.  In one she had R+2N vs. R when she was in the second grade.  I'd told her to exchange pieces when she was ahead in the endgame.  She exchanged rooks and then it was a draw!  This was before the 6-piece tablebases, but I played it against the computer, and it seemed pretty easy for the R+2N to win.  Bubba73 You talkin' to me? 02:30, 27 February 2019 (UTC)


 * And in the very next tournament, she got to 2N vs. P. Of course, winning that was way above her ability.  But she did figure out that she had to give up a knight to keep the pawn from queening.  I was proud of her for that.  Bubba73 You talkin' to me? 02:33, 27 February 2019 (UTC)
 * Your daughter certainly gets into weird situations at the board! I haven't had any in person either. Even among basic checkmates, I don't remember ever having to do K+2B vs K or K+B+N vs K in a real game, although I do know how to do them. (I don't even think the latter is as hard as its reputation, and I'm not that good a chessplayer. Two knights versus pawn – now that is hard in my experience trying it against a computer. I know that generally I'm supposed to park a knight in front of the pawn and then confine the enemy king with my king and knight but actually doing it is much harder than it looks.) Telling my computer to play against itself does seem to produce R+B vs R with some small frequency, and the result seems to confirm the statement I saw somewhere that positions in this endgame that come from normal games tend to be drawn. Other than Q vs R I don't remember seeing any others show up more than once. I don't remember what the 7-man endgame was because that one happened around a decade ago, but it was a 50-move rule draw with about equal if unbalanced material IIRC. (Even if it had been QN vs RBN I'm sure computers won't find the win without having 7-man tablebases installed, and it would probably run afoul of the 50-move rule anyway.) But I'm sure someone has better statistics that are more than OR (and bad OR at that since I lost many of those games in a computer crash a while back). Double sharp (talk) 03:12, 27 February 2019 (UTC)


 * These generally don't consider the 50-move rule. I've had the 2B checkmate.  Once in a tournament I had B+N+P vs. a minor piece (I don't remember which one).  I was playing along - then I realized that if he sacrifices the piece for the pawn, I have to try the B+N checkmate, which I doubted that I could do.  I protected the pawn and promoted it.  Bubba73 You talkin' to me? 05:10, 27 February 2019 (UTC)


 * I've had some interesting ones in practice, not pawnless. One I drew R+P vs. Q.  I sacrificed some material to get to it (a minor piece for a pawn, I think), because I was pretty sure I could draw it.  I've also drawn some wrong rook pawn/wrong bishop ones, usually maneuvering to get into the position.  I told my daughter that if she would let me teach her more about endgames (not just these unusual ones), she could get an extra a half point every 10-15 games (but she wouldn't listen).  Bubba73 You talkin' to me? 08:03, 27 February 2019 (UTC)
 * I've always been very fascinated myself with the cold precision and deceptive simplicity of endgames; OTOH, most of the ones that come up in games have pawns in them, so while Stiller's monsters are awe-inspiring, they're not quite relevant to the average chessplayer as anything other than a source of awe. And obviously the reason why most endgames in real games have pawns is because we start with more pawns than any other piece type.
 * P.S. After retrying the B+N mate against a computer just now, I'll add a clarification: it's easy when you know the technique, but it's definitely nothing like as easy if you've half-forgotten it because you haven't practiced it for about three years, so I guess it actually ought to be considered hard and it only feels simple once you learn it and the hard work is over. ^_^ I think I'm starting to get it again, although I'm not confident that I'd be able to do it in tournament conditions yet. (Maybe that is one of the biggest reasons why it has its reputation – its perceived difficulty is increased because you rarely have the opportunity to do it. No doubt most pawnless endgames would have this problem to an even greater extent: I've been trying Q vs R against a computer too with no success yet. Stiller's monsters at the very least show us how far removed the needed moves can feel from ordinary chess logic, so for these real difficulty is piled up on difficulty due to the lack of regular reinforcement!) Double sharp (talk) 15:00, 2 March 2019 (UTC)

Extra rooks?
Sticking to 7 men as a sort of maximum (since beyond that we lack tablebases), it seems odd that we include adding an extra pair of minors to R+minor vs R but not an extra pair of rooks. Or is the result less interesting? Double sharp (talk) 02:18, 27 February 2019 (UTC)


 * I think it sould be included. When I collected this stuff, it was all (or almost all) from books, and before seven-piece tablebases.  In fact, IIRC, at the time of the first edition of Nunn's book on the subject, the six-piece endgames had not been finished, and he added some of that to the second edition.  I haven't kept up with the tablebase results.  Bubba73 You talkin' to me? 02:25, 27 February 2019 (UTC)
 * I haven't really kept up with them either, and based on the above talk page sections, getting the data from the Lomonosov tablebases would require some kind of way to make sure instant or near-instant conversions are excluded. Incidentally not all the cases of R+2 minors vs R+minor are actually mentioned. Among the missing ones, I would guess that RBB vs RN is won (as it is won even if the rooks are exchanged), while I would guess that RNN vs RB or RNN vs RN would be general draws (though RNN vs R is generally won despite the need to avoid the even exchange, I wouldn't guess that the power of the R+2N to drum up a mating attack is enough against a rook and a minor piece). But these are just my guesses.
 * In general I think that it would be ideal to include results for all the 7-man endgames, with the obvious exceptions of 6 vs 1 and 5 vs 2 that aren't Q vs 4 minors as their results are really obvious. ^_^ But I don't really have access to the best sources to make that happen. Double sharp (talk) 02:29, 27 February 2019 (UTC)
 * Aha! Entry 298 of Tim Krabbé's open chess diary covers this. RRN vs RR and RRB vs RR are both general draws, but some very long wins exist. Time for me to add it, then. Double sharp (talk) 02:39, 27 February 2019 (UTC)
 * ✅ Double sharp (talk) 02:45, 27 February 2019 (UTC)

Surprises from the aforementioned post by Marc Bourzutschky

 * 3 minors vs 2 minors, in normal situations, is a win exactly when the stronger side has the bishop pair and the weaker side lacks it.
 * If there are same-coloured bishop pairs, weird things can happen. BBN (different colours) vs BB (same colours) is not generally won, but has very long winning lines (220 moves to capture).
 * R + 2 minor vs R + 1 minor is general win, agreeing with what Speelman reported as the thought of 1981. Apparently this was surprising when found out, though.
 * 2R + minor vs 2R is not generally won, but is won more often than was thought.
 * When queens are around, tempo is more important than material. QQ vs QQN is won for the weaker side half the time, when he has the move!!

Double sharp (talk) 07:40, 3 March 2021 (UTC)


 * BBBB (two pairs) vs RR is general draw. But a surprisingly complicated one! Double sharp (talk) 07:37, 5 March 2021 (UTC)


 * Nice. Can you incorporate these into the article? Bubba73 You talkin' to me? 07:44, 5 March 2021 (UTC)
 * ✅ Double sharp (talk) 07:48, 6 March 2021 (UTC)

The Rybka Forum has been closed, so hopefully nobody objects if I reproduce the post as a quote:

In the endings with 3 minor pieces vs. 2 minor pieces the only generally won cases that don't involve one side having more than one bishop of the same color seem to be kbbnknn and kbbnkbn. kbbnknn is no big surprise, given that kbbkn is a general win. There are some bizarre endings with same colored bishops. For example, kbbnkbb where Black has two bishops of the same color (while White has opposite colored bishops) contains winning lines of 220 moves to capture, even though the ending does not seem to be a general win.

However, rook + 2 minor pieces vs rook + minor piece seems to be a general win, one of the bigger surprises that has come out of 7-man databases.

2 rooks + minor piece vs 2 rooks seems to be close to a general win, at least it is won more often than most people expected.

Pawnless endings involving queens with one side having an extra minor piece don't seem to be general wins. In those endings, having the move is often more important than the extra piece, for example white to move in kqqkqqn wins about 50% of tje time. Double sharp (talk) 11:00, 15 November 2021 (UTC)

What about two rooks vs rook and minor piece endgame?
Unless I'm missing something, it's not mentioned anywhere in article. 83.29.16.137 (talk) 14:23, 23 August 2023 (UTC)
 * You are right.
 * That said, R vs minor is usually a draw, and an extra pair of rooks helps the side down the exchange. Therefore, probably RR vs R+minor is usually drawn too. Double sharp (talk) 05:45, 25 September 2023 (UTC)