Game #311: beating a 1159 with the Giuoco Pianissimo as Black
On March 24, 2024, in game #311 of my rapid journey, I beat kraziianmol (1159) with Black while rated 996, a gap of 163 points. The opening was the Giuoco Pianissimo, the slow Italian with d3, and as Black I got it wrong on move 4 and much worse on move 5. The engine had me lost at +5.32 before move 6. Then my opponent missed the win, and I played the rest of the game better than almost any game I'd played that month.
It's one of the ten biggest upsets of my first 2,116 games, and I'm writing it up because it shows two things at once: a real hole in how I played the Giuoco Pianissimo as Black, and what happens when the opponent doesn't punish it.
Where I was that week
This game is in week 11 of the journey, a short and bad week: 13 games, 4 wins, 9 losses, rating 1022 → 999. I'd lost my two previous games, to capitalbrae (991) and javierhcc (1047), and come in at 984. The win took me to 996. The next game I lost again, to mrchessjumbie (1095), and I ended the week just under 1000.
| Game | #311, March 24, 2024 |
| Opponent | kraziianmol (1159) |
| My colour, rating after | Black, 996 (from 984) |
| Time control | 20 minutes, no increment |
| Opening | Giuoco Piano Game, Giuoco Pianissimo Variation (C50) |
| Result | Win, White resigned after move 21 |
| Accuracy (Stockfish) | Me 88.9, White 71.7 |
| My blunders / mistakes / inaccuracies | 1 / 1 / 1 |
Move 4: the knight on e7
The game started 1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.d3, the quiet Pianissimo setup. I'd played the Pianissimo as Black eight times before in this journey, with four wins and four losses. On move 4 I developed my king's knight to e7.
The engine marked it as a mistake (+0.13 → +1.59). The knight on e7 doesn't control the squares a knight on f6 would: from f6 it guards h7 and h5, and from e7 it guards neither. White saw that at once: 5.Ng5, attacking f7 with knight and bishop together.
Move 5: castling into the attack
This is the move I'd most like back. With White's knight on g5 and bishop on c4 both aimed at f7, I castled.
The engine scored castling as a blunder: +1.79 → +5.32. Its line for White starts with the queen to h5, attacking h7 with the knight on g5 supporting it and threatening checkmate on the next move. Black can stop the mate, but then the knight takes on f7 and the material goes. The engine's move for me was d5, the pawn hitting the bishop on c4 and cutting its diagonal.
Move 6: White's miss, and mine
White didn't play the queen move. They pushed the f-pawn to f4 instead, and the engine called that a blunder too: the evaluation swung from +5.32 to -2.00, better for me. That was the turning point of the game, and I didn't make it. My opponent did.
I played h6, attacking the knight. The game analysis marked it an inaccuracy; the engine preferred d5, the same idea as the move before. Even so, my move kept me better. White took on f7 with the knight, I took back with the rook, the bishop took my rook with check, and my king took the bishop. After 9.Qh5+ Kg8 White had given two minor pieces for a rook and a pawn, and the engine had me at -1.77.
The middlegame: 96.7 accuracy
From here on I played cleanly, and I don't say that often in this series. My accuracy in the middlegame was 96.7, with no mistakes and no inaccuracies. White's f-pawn took on e5, I hit it with d6, White took on d6, I recaptured with my queen, and my knight jumped to d4. On move 13 White developed the knight to a3; I took it with my bishop. White answered with a bishop sacrifice on h6, a mistake by the engine's count (-3.99 → -6.66), and my knight took on c2 with check, forking the king and the rook on a1.
One move later I had a choice that looked risky.
My bishop went to g4 with check, attacking White's queen on h5 at the same time. White took it with the queen, and my queen then took the bishop on h6. The game analysis had preferred taking on h6 straight away; when I checked the position again for this post, the engine put Bg4+ first, with both moves completely winning. The difference between them doesn't matter. What mattered was that every capture was counted before I played it.
On move 17 White's rook came to f6 and attacked my queen. My queen took it. On move 19 my knight came back to d4 with check, on move 20 my queen took the pawn on g2, and after 21.Qf2 White resigned. I was two knights and a bishop up against one extra pawn, with 11 minutes 43 seconds left on my clock.
What the engine says about this upset
Here's the honest version. I made three errors in this game by the engine's count, all in the first six moves: the knight move, the castling blunder and h6. White made four, including the one that mattered most, 6.f4 instead of 6.Qh5. If White had found the queen move, this would have been a short loss, not an upset.
But I also don't want to undersell the rest. My middlegame was 96.7 and the engine flagged nothing I did after move 6. At this rating, I had plenty of games where I got a winning position and lost it. This one I got and kept.
The lesson I took from game #311
The Giuoco Pianissimo as Black is quiet only until White's knight reaches g5. Before I put my king's knight anywhere other than f6, I now check whether g5 and h5 become free squares for White. And when the knight and bishop are both aimed at f7, castling is not a safety move: it puts the king where the attack is going. The pawn to d5 was the engine's answer twice in a row, and it's worth remembering why: it hits the bishop and cuts the line to f7 in one move.
What came next
The lesson didn't stick right away. Seventeen days later, in game #361 against SikkenBerend (1172), I played the same Pianissimo as Black, put my knight on e7 again, and the engine flagged it again. That game was part of my 11-game losing streak. The week after this one is week 12, and the full story is on the journey page. You can replay game #311 on Chess.com.
This is the kind of opening hole that is easy to miss game by game and obvious across many games. Finding the moment and naming the reason in plain words is what I built Chess Coach for, because Chess.com's review doesn't always explain the moves. Here's how the analysis works.
Frequently asked questions
What went wrong for me in the Giuoco Pianissimo as Black?
My fourth move. I put the king's knight on e7 instead of f6, which the engine marked as a mistake, and after White's knight jumped to g5 I castled into a mating threat. By move 5 the engine had White at +5.32.
How did a 996 beat a 1159 in this game?
White missed the queen move to h5 that was winning and pushed the f-pawn instead, a blunder that swung the evaluation to -2.00 in my favour. From there I made no more mistakes: my middlegame accuracy was 96.7 and White resigned after move 21.
What should Black play against Ng5 in the Italian?
In this game, with my knight already on e7, the engine's move was the pawn to d5, hitting White's bishop on c4 and blocking its diagonal to f7. Castling, which I played, left the h7 and f7 squares open to White's queen and knight.