The costliest move in chess

Intro

I tried to see what are the most common lines in the Lichess database, at different ratings, time controls, etc. I was quite surprised to see that there are just a few variations that are played the vast majority of times. Since this post is not about those specific statistics, I will just summarize the results:

Fried Liver Attack

• from the Italian Game, Two Knights Defense, Knight Attack
• played by players under rating 1400, regardless of time controls
• played the most in classical and correspondence, regardless of rating

Greco Gambit

• from the Italian Game, Classical Variation
• played by players 1400-2000
• played the most in blitz and rapid

Polerio Defense

• from the Italian Game, Two Knights Defense, Knight Attack
• played the most in bullet

Chelyabinsk Sicilian

• from the Open Sicilian Defense: Lasker-Pelikan Variation, Sveshnikov Variation
• played by players rated 2000-2200

Najdorf Sicilian

• from the Open Sicilian Defense: Modern Variation
• played by players rated 2200+

Najdorf Sicilian - English Attack

• from the Open Sicilian Defense: Modern Variation
• played by the Masters

The result is kind of funny for me, because I was never so low rated as to care about or even encounter the Knight Attack, nor so high rated as to deal with advanced Sicilians and I never heard about the Greco Gambit.

Yet, it gave me this idea. If these are the most played variations by far, does it mean that the blunders that lose in these common lines are the costliest chess mistakes?

Well, let's see!

Knight Attack

The following position in the Fried Liver attack has been reached in 3.5 million games.

White sacced the knight, checked the king. There are two main moves played from this position: Ke6, protecting the knight (average rating 1519), and Ke8, running to safety (average rating 1368). The second move is a mistake, losing 2 eval points, but what comes next is even worse, because after 7...Ke8 8. Bxd5 the most played move is 8... Nd4 (average rating 1327), hanging mate in one! Indeed, in the main line for under 1400s the game ends with a mate on the 9th move.

Here is an example game, but take heed, this exact game has been played 316,819 times!

But what about playing Ke6? Well, the most played moves lead to this position:

The king protected the central knight, White added a new attacker with Nc3, Black defended and attacked with Nb4, White castled, Black protected the central knight again with c6, then White played d4. Evaluation is +0.9 if Black plays Qf6. Yet Black plays Nxc2 46% of the time (average rating 1660).
Eval goes to +3.3, but after dxe5 the move played 80% of the time is Nxa1, which leads to a +8 eval! That position has been reached 8664 times.

OK, 8664 is a lot less than 316,819, but this is the position reached most after Ke6.

Here are some example mates:

The first has been played 46 times and the second 109 times.

OK, what about the Polerio Defense? What is that about? Well, 5... Na5 is the best move in the position. It attacks the light square bishop and thus prevents the Fried Liver and forces the bishop to move. This is where it gets funny:

From this position, the position we reached as the most common bullet line, the main move is Qf3. But, if you look at ALL the games reaching this position, the most played move is 8.Ba4 which is a -2 eval mistake! Average rating for those making this move: 1596. However, the line to take advantage of the mistake is not the main line. Most of the time Black doesn't know how to capitalize on this trap.

Here is a game where Black does it right. The important move is 10...Be6 and NOT 10...Be7 which is much more common:

Anyway, let's get back to the main bullet line. It doesn't actually end in any terrible situation. In fact, it goes relatively equal until move 14:

Only 689 games reached this position and of them all, 394 make the move 13...Nxe4 which is a mistake pushing the eval at +1.33. I'll give you a game, but to be honest, it just seems relatively equal (win rate for White is just 54%):

Funny enough, it's a rapid game.

Greco Gambit

The Greco Gambit is defined by the following position on Lichess:

As you can see, no one gambits anything! It's getting even funnier when, depending on the moves, you go into "Greco's Attack", "Greco Gambit, Greco Variation" or "Greco Gambit, Main Line"

Most of the variations don't go anywhere soon, but here is a very common position:

The best move in the position is taking with the bishop, which happens 50% of the time, but 41% of the time Nxc3 is played by players rated 1644 on average, which naturally leads to the following continuation: 9. bxc3 Bxc3? 10. Qb3 Bxa1??

The trigger position happened 436,354 times and the position after the continuation above has been reached 102,768 times.

Here is a game:

Najdorf and Chelyabinsk Sicilian

Here you probably expected a total chess noob to tell you about the secret blunder every 2000+ player makes in the Sicilian. Well, I appreciate the confidence, but it's not going to happen. People over 2000 rating know their chess and play their lines fully booked up.

While I've identified the "most common lines" and the blunders in it, those usually happen in a few games and cannot be generalized.

Conclusion

I've identified two positions, one from the Knight Attack in the Italian Game, where more than 300,000 games ended in mate on the ninth move after getting an advantage in a position reached 1,200,000+ times, and the second from the Greco Attack in the Italian Game, where more than 100,000 games ended in a vastly superior position for White leading to mate.

I would say that all of these positions benefit from further analysis, but the clear winner for the costliest chess mistake (statistically speaking) is ...

[FEN "r1bq1b1r/ppp2kpp/2n5/3np3/2B5/5Q2/PPPP1PPP/RNB1K2R b KQ - 1 7"]
[SetUp "1"]
7... Ke8

reached 1,289,167 times and leading to a White win in 81% of the cases, which is a little over a million games, one in every 5000 chess games played on Lichess.

Hope this was as fun for you as it was for me.