Game #396: how to beat the Grob Opening as Black, from a 980 against a 1149
On April 21, 2024, in game #396 of my rapid journey, I beat wormy84 (1149) with Black while rated 968, in 19 moves. White opened with 1.g4, the Grob Opening, and the way I learned how to beat the Grob Opening as Black in that game was simple: don't do anything clever, develop, and wait for White's pawn pushes to open a road to their own king. My rating went to 980, a gap of 169 points to my opponent. By that gap, it's the fifth-biggest upset in my first 2,116 games.
It was also the first time anyone played the Grob against me in this journey. I've faced it three more times since.
Where I was that week
This game sits in week 15 of the journey, a week where I went 3–9 with Black and fell from 993 to 976. It came right after the week of my eleven-game losing streak. The two games right before this one were both losses with Black, to happyhappyabby (1056) and Anya_Kosmos (995), with my accuracy at 47.2 and 51.9. So I came into game #396 on two bad games, against a player 181 points above me, and it turned out to be one of my three wins with Black that week, without a single blunder.
| Game | #396, April 21, 2024 |
| Opponent | wormy84 (1149) |
| My colour, rating after | Black, 980 (from 968) |
| Time control | 20 minutes, no increment |
| Opening | Grob Opening (A00) |
| Result | Win, White resigned after move 19 |
| Accuracy (Stockfish) | Me 89.2, White 80.3 |
| My blunders / mistakes / inaccuracies | 0 / 1 / 3 |
The whole game took about six minutes of real time.
The Grob, and why I didn't try to refute it
The Grob is 1.g4: White pushes the knight's pawn two squares on the very first move. It looks aggressive, but the engine doesn't like it. Stockfish called 1.g4 a mistake and rated the position -1.36 right away (evaluations are from White's side, so minus means better for Black). My answer was 1...d5, a pawn to the center, and it was the engine's first choice too. That's the part of the game the analysis marked as good.
White kept pushing: 2.g5, then the bishop to g2 aimed at my d5 pawn, then 4.h4. I put my bishop on f5 and a pawn on e6 to support d5, and when White's g-pawn sat on g5 I challenged it with h6. White took, and I took back on h6 with my knight. By move 5, White had no g-pawn left at all, and the king was still in the middle on e1.
Moment 1: the pawn push that opened the diagonal
On move 6 White pushed the f-pawn one square. The engine called it a blunder: the evaluation went from -1.40 to -3.72. I can't see every reason from the board alone, but one is visible: with the f-pawn gone from f2, the diagonal from h4 to White's king on e1 was open. The engine's suggested line for me even shows a bishop check on h4 a few moves later. That diagonal is the whole story of this game.
I didn't use it straight away. On move 7 I gave a bishop check from b4 instead, and the engine was not impressed.
The check itself was rated an inaccuracy (-4.15 to -3.01), and after White blocked with the c-pawn, I sent the bishop to a5 instead of back to e7. That was my only mistake of the game: -3.20 to -1.59, half of my advantage gone. White then chased the bishop with b4 and the knight with b5, and my pieces ended up on a5 and b6, on the queenside. With hindsight, the engine's bishop to e7 points towards h4, the end of the diagonal White had just opened.
Moment 2: the queen capture I passed on
White's pawn pushes kept costing them. After 11.e3 (a mistake by the engine's count) and 12.a4, I castled queenside, and on move 13 White pushed the e-pawn again, to e4. I took it, White took back, my bishop came to g4, and White put the queen on c2.
Queen takes on d3 was a double attack (one move that threatens two things at once): it hit White's queen on c2, which had no defender, and the knight on b1 behind it. The engine rated it -6.83. I played queen to d6, which still kept me at -4.69, a big advantage, but the analysis marked it as an inaccuracy: I passed on the most forcing move on the board.
Moment 3: the check on g3
On move 16 White moved the bishop from g2 to f3, which the engine called a mistake (-9.87). Now my queen on d6 had a straight diagonal to g3, with White's king still on e1.
The game analysis preferred bishop takes bishop. When I checked the position again for this post, the engine's first choice was my move, the queen check on g3, at -11.24, and it described it as a check that attacks the knight on g1 and the bishop on f3 at the same time. Either way, this is where the f-pawn push from move 6 came back: my queen reached g3 through the open squares in front of White's king.
Moment 4: the queen that couldn't block
White blocked the check with the queen on f2. The problem was my bishop on b6, the one I'd been chased to the queenside with. The diagonal from b6 to f2 was open all the way, because White's e-pawn had left e3 on move 13. The engine described White's move as attacking my queen and my bishop, "but the queen can now be captured". I took it with the bishop, with check.
The engine actually preferred queen takes queen with check (-21.94 against my -17.47), but by then both were "completely winning". White's king went to d2, my bishop took the knight on g1, and after White's bishop took mine on g4, the engine found a forced checkmate for me.
The line was queen takes on d3 with check, queen to g3 with check, and queen to f2, checkmate. wormy84 resigned here, and I didn't have to find it.
How to beat the Grob Opening as Black, from this game
I don't think I beat the Grob with opening knowledge. I didn't know any lines against it. What worked was a short list:
- Take the center first. 1...d5 was the engine's move and kept everything simple.
- Develop, don't chase. My worst moves in this game were bishop checks and bishop walks to the edge. The engine's moves were quiet developing moves like bishop to e7.
- Watch the diagonals the pawns open. Every pawn the Grob player pushes on the king's side opens a line. f2–f3 opened the diagonal to the king, and e3–e4 opened the b6–f2 diagonal that won the queen.
- Look for the forcing capture. On move 15 I chose a slower queen move over a capture that attacked two things. It didn't cost me here; against a more careful opponent it might have.
What came next
The very next game, #397, I lost with White to Sbarrini1968 (1159), and I finished the week at 976. The following week, week 16, went better: 7 wins, 7 losses and 976 → 997. The next time I faced the Grob, in game #515 against oceano929 (1087), I lost. So this game didn't mean I'd solved the opening; it meant that on that evening I kept it simple and my opponent didn't. The week before this one is week 14, and the whole series lives on the journey page.
If you'd like your own games broken down like this, with the moment, the better move and plain words for why, that's what Chess Coach does inside Claude: here's whether Claude can analyze chess games and how the analysis works. You can replay game #396 on Chess.com.
Frequently asked questions
What is a good first reply to the Grob Opening?
In my game, after 1.g4 I played 1...d5, which was also the engine's first choice. Stockfish already rated the position -1.36 after White's first move, so the Grob gave Black an edge from move one.
How did a 980 beat a 1149 with the Grob on the board?
White kept pushing pawns: g4, g5, h4, f3, b4, b5, a4. The engine called the f-pawn push on move 6 a blunder, because it opened the diagonal towards White's king. My queen later used that diagonal, and White's queen move on move 17 lost the queen to my bishop.
Did you play the Grob game perfectly?
No. Stockfish gave me 89.2 accuracy, no blunders, but one mistake and three inaccuracies. My bishop's walk on moves 7 and 8 gave back about half of my advantage, and on move 15 I missed a queen capture that attacked two things at once.