One of the many great things about Lichess is the database and opening explorer. With them one can look into which openings are being played at different levels and how well they score.

I got interested to see who well different opening moves score at different levels. There are an endless number of different openings so I decided to mainly limit myself to the first few moves of the game.

Scores for first moves

The first thing I wanted to look at is the score of different first moves for White. Across the different rating ranges, White scores around 52% and this remains roughly constant. So if the score for White is above 52% (0.52 in the graphs) the opening is better than average for White.

I was surprised to see that there was little difference between the most common first moves for White. I would have guessed that less common options would score better for White at lower levels as the opponents aren’t used to facing these openings. But the difference in score is not too large between the different options.

When looking at different responses to 1.e4 the picture changes and it looks like Black scores better when they choose less common moves.

At lower levels, White scores much better in 1.e4 e5 than any other common response to e4. There are even some openings where White scores below 50% meaning that Black has better practical winning chances after just one move.

But as one moves up the rating range, the scores of the different responses start to converge. The Sicilian for example changes from being an opening that scores better for Black to the second best opening from White’s point of view. It’s interesting to note that the Modern (1.e4 g6) does very well for Black, even at the highest Lichess ratings.

I also took a look at different responses to 1.d4.

The scores don’t shift as dramatically as in 1.e4 openings but there are still some interesting developments.

Firstly, White scores extremely well in the Queen’s Gambit Accepted at lower levels with over 56% but as the strength of the players increases, it actually turns out to be the worst scoring opening for White I looked at.

Secondly, the change in score of the Grünfeld is really interesting. At the lowest levels, it scores similarly to the other openings but in the 1400-1800 range it scores very well for Black before returning back to an average score. I’d guess that in the 1400-1800 range, the imbalance in opening knowledge in the Grünfeld is the largest: at lower levels both players know little about the opening and at higher levels both players know a lot. In between Black knows significantly more about the opening than White and the Grünfeld is an opening where this knowledge transfers to an improved score.

Share

Gambits

Finally, I thought that it would be interesting to look at different gambits.

It’s interesting to see that from a score perspective, the Englund, Budapest and Smith-Morra Gambit score roughly the same as the average opening. The King’s Gambit does very well below the 1800 level and then scores in line with the other gambits.

The Marshall Attack and Benko Gambit both show a similar shape to the Grünfeld Defense, where Black scores very well with them in the 1400-1800 range. Again, I think that the gap in opening knowledge is highest at that range. And as these are sounder openings compared to other gambits I looked at, Black can use the knowledge advantage to actually score much better than White.

Conclusion

The U-shape of openings like the Benko Gambit and Grünfeld really stood out to me. I never thought about it before, but it makes sense that double-edged openings that are objectively balanced favour players who know significantly more about them. And 1400-1800 seems to be the rating range where one player knows the lines they play well while their opponent might not be as well prepared against side lines.

Let me know if you have a different explanation for the U-shape of some openings and if there are other opening stats you’re interested in.

Leave a comment