White's Best First Move

BOBBY Fischer famously observed that 1.e4 is "best by test."

He first made the remark in his December 1963 Fischer Talks Chess column in the US federation's magazine Chess Life.

Fischer made the assessment based on his experience and judgment - there was little chance in those days of backing up such an assertion with meaningful statistics.

Today the situation is different, as ChessBase's 2025 Mega database reminds us.

I have used it discover what the statistics say about White's best move in modern chess.

I called up games played in the last five years - a massive 2,212,990 of them.

They include players of all abilities, but it is probably fair to say such a collection is skewed to above-average chess in that casual games and many club matches are excluded.

The first surprise is that in those games White scores only 53%, whereas it is often stated that in large databases White scores 55%, or, at worst, 54%.

In those games White had an average rating of 2122, returning an average performance of 2133 - again a lower figure than I expected.

It turns out that in modern chess, as defined above, 1.e4 scores just 52%, while 1.d4 pips it with 53%, but both moves lag behind 1.c4 (55%) and 1.Nf3 (56%).

Also doing relatively well is 1.b3 (56%), a move Fischer dabbled with, but all the moves pale in comparison with 1.h4 (60%), 1.a3 (61%) and 1.a4 (62%).

By now alarm bells should be ringing - clearly raw percentages are not enough to go on.

Luckily chess, unlike most sports, has a well-developed rating system that should give more-informative results.

For example 1.a4, which scores 62%, gives a rating performance of -63. In other words, whites performed at 63 elo below their average rating.

Also doing badly is the other flanking pawn thrust, 1.h4, whose 60% score comes at the price of a -7 rating performance.

In both cases whites must have been on average quite a bit higher rated than the average black.

The move 1.a3 (61%) bucks the trend, at least to an extent, as whites return a +7 rating performance. This is better than the -1 of the 56%-scoring 1.b3 and the +2 of the also 56%-scoring 1.Nf3.

But the three traditional pawn thrusts do best, with 1.c4 (55%) performing at +17, 1.d4 (53%) at +19 and 1.e4 (52%) at +21.

It seems Fischer may have been right all along.