Double pawn sacrifice

Mid Journey AI created

AlphaZero's Mind-Blowing Chess Strategy: Double Pawn Sacrifice and Queen Dance

ChessAnalysisOpeningTacticsChess engine
Alphazero makes use of the "Polugaevsky Gambit" in the Queens Indian and adds another pawn sacrifice

Hi all

I have revisited the remarkable Game 5 of the 2017 match between Alphazero and Stockfish. It features a remarkable double pawn sacrifice which creates very interesting dynamic play that ends up winning the material back with interest.

The game starts with a Queen's Indian Defence

This is the "Polugaevsky gambit" in the Queens Indian defence. It is named after him partly because of the following amazing game he played vs Korchnoi:

In the Alphazero game, we see a fascinating continuation. One early key point is to get a Knight onto f5. Kasparov has also made use of this gambit with great effect.

Here comes a 2nd pawn sacrifice:

Very "dynamic chess"

- 2 pawns down but with rather a lot of compensation including Black's awkward Queenside pieces, and White's dark square bishop without a counterpart:

A stunning Bg5:

A concession had to be made later giving up the exchange - c5

Winning the exchange

Final position

Video annotation

Key takeaway points

  • The gambit is very dangerous as it is
  • Adding a second pawn sacrifice makes the gambit even more dangerous - supercharging it
  • Stockfish had to give back material when it realised the compensation was so powerful
  • Dynamic chess is alive and kicking - perhaps we have been trusting our engines too much before the Neural Network wave of engines

I hope you enjoyed this blog :). Any likes and follows are really appreciated. Also, I also have some interesting chess courses at to check out.

Cheers, K