22 January 2018

Checkmate Data

It is commonplace to speak and write about checkmate patterns. Certain recurring checkmates happen over and over again in games no matter the skill level of the players. Of course, once players rise above a certain level, checkmates become rare due to resignation either due to a checkmate threat or one player gaining an overwhelming material advantage.

Do claims about checkmate patterns stand up to data from played games? It is a simple, albeit time consuming matter to examine every game in the database that ended in checkmate. It may be possible to name every pattern in these checkmates and then list them in order of frequency.

Running on my desktop, ChessBase was able to find every game in my largest database ending in checkmate in about one minute. As I prefer quality over quantity, my largest database is smaller than many others, containing a mere 5.9 million games.

In this database, 148,127 games end in checkmate.

The last piece to move was most often the queen, and presumably this piece was always giving the check that is checkmate in all of them. Queen moves account for 59% (87,894) of the games in this selection.

The rook is the last piece to move in 24% (35,473) of the games. Most often, the rook is likely giving check and mate. However, a rook move can produce a discovered check by bishop or queen.

A little more than 7% (10,990) of the games ending in checkmate involve a knight making the last move of the game. Some of these, no doubt will be discoveries, but it's a reasonable prediction that the knight is checking the enemy king in most of them.

A bishop moves to checkmate the opponent in 6% (8,898) of the games.

Pawn moves account for slightly more than 3% (4,847) of the games. It will be interesting to study how often the pawn itself delivers the decisive check.

King moves account for a mere 25 games. More than half of these are by castling, so a rook is giving checkmate. The king cannot deliver check, but it can cut off the escape of an enemy king.

For example, in this game between two youth players from 2005, White's king moved so as to check by the rook and cover one of the escape squares.

White to move

One of the games that ended with O-O# could have gone on two moves longer with better defense. The combination strikes me as instructive. It is from Suess -- Hurme 1969, a game played as part of team world championship qualifier.

White to move


  1. That's an interesting mate. As you say, instructive. I am sure I have never seem mate with 0-0 before.

    I admit I tried at least one alternative, as the obvious move seemed not to work, but you can get there. I also used a board.

    1. Checkmate by castling is rare, but I've found at least four games that end O-O# and four more that end O-O-O#.

  2. 2nd problem:

    1.Qd1 Q(any), 2.Bc4 e6, 3.Qh8+ Kf7, 4.Qxf6+ Kg8, 5.Rh8 mate, or at least that's how I'd play it.

    1. Something's not right. White's queen cannot move to d1.

  3. 1.Qh8+ Kf7 2.Ng5+ fxg5 3.Bc4+ e6 (3...Be6 is the same) 4.Rh7+ Nxh7 5.Qxh7+ Kf8 6.0-0#

    I once had a cool game where 0-0-0 was the final move: