Week 105: kingside fianchetto weaknesses, in one game
One game this week, and it was a loss. On February 11, 2026, after 18 days without a rapid game (my previous one was on January 24), I played a single rapid game as Black and my rating went from 1261 to 1260 (−1). It's a small entry, but the game is a clean lesson in kingside fianchetto weaknesses: what happens to a king on g8 when the bishop on g7, the piece that usually guards it, gets traded away.
The week in numbers
| Games | 1 |
| Wins / losses / draws | 0 / 1 / 0 |
| Rating | 1261 → 1260 (−1) |
| Average accuracy | 79.2 % |
| Blunders | 1 |
| Opening | Modern Defense with 1 e4 |
My opponent was dedialimahmudi42gmailco (1290), and the engine summary is honest: 87.9 % accuracy in my first twelve moves, 47.0 % in the two middlegame moves that followed. White played 93.8 %. The whole game lasted 15 moves; Chess.com records it as abandoned after White's 15.Ng5 (game).
How the game went
White chose an unusual start: 1.e4 g6 2.Bb5, then Nc3, d3, Nf3 and a clear plan: the bishop to g5 and the queen to d2. That pair aims at h6: the bishop goes there to trade itself for my g7 bishop, and the queen recaptures right next to my king. On move 9 the engine wanted me to stop it with 9...h6, attacking the bishop on g5 (−0.66, slightly better for me). I castled instead, 9...O-O, and the engine's lines after that already include Bh6 for White.
Move 12: the pawn push that opened h6
12...h6 attacks the bishop, which feels natural. But with my king already on g8 the pawn on h6 can be taken: the engine's reply is 13.Bxh6, a bishop that "takes the pawn protecting its king" and attacks my bishop on g7 with the rook behind it (+2.13, clearly better for White). That is exactly what happened: 13.Bxh6 Bxh6 14.Qxh6. The calm 12...Nac6, bringing my knight back from a5, kept the game about equal (+0.08). The same pawn move that was right on move 9 was wrong on move 12, because in between I had castled into its range.
Move 14: the move that allowed mate
Now the fianchetto is gone. My g7 bishop has been traded, White's queen stands on h6, and my h-pawn, which used to stand on h7, is gone. I played 14...Bc6, a quiet move on the other side of the board. The engine's answer is 15.Ng5, a knight move that "threatens checkmate on the next move", and the line is forced: 15...Re8 16.Qh7+ Kf8 17.Qxf7# (mate in three). With 14...f6 the pawn controls g5, and White is clearly better but there's no mate (+2.29). My opponent played 15.Ng5 and the game ended there.
What the engine's numbers say
The game was decided by two moves in a row: 12...h6 (rated an error) and 14...Bc6 (the blunder). Before move 12 my accuracy was fine and the evaluation stayed within one pawn either way. This is the pattern that makes the Modern and Pirc risky for me: I spend moves on the queenside (…a6, …b5, …Na5, …c5 here), while White's plan on the kingside needs only three moves — Bg5, Qd2 and Bh6.
The lesson of the week
When I fianchetto, the bishop on g7 is part of my king's shelter. If White lines up the queen and bishop against h6 (Qd2 and Bg5 or Be3), I either prevent the trade or prepare for life without the bishop. And once the bishop is gone, my first question every move is: can White's queen and knight reach h7? Here 14...f6 answered it; 14...Bc6 ignored it.
All games this week
| Game # | Date | Colour | Opponent (rating) | Result | Opening | Accuracy | Link |
|---|---|---|---|---|---|---|---|
| 1977 | 2026-02-11 | Black | dedialimahmudi42gmailco (1290) | Loss | Modern Defense with 1 e4 | 79.2 | game |
Previous: week 104 · Next: week 106 · Back to the journey · Want the same review for your games? See how the AI chess analysis works
Frequently asked questions
What are the weaknesses of a kingside fianchetto?
When the bishop on g7 is traded, the dark squares next to the king, f6 and h6, lose their main defender. In game 1977 White traded it with 13.Bxh6 Bxh6 14.Qxh6, and their queen landed two steps from my king.
How do you defend against Qh6 and Ng5?
Cover h7 before the knight arrives. In game 1977 the engine's defence was 14...f6, which kept White only clearly better (+2.29); my 14...Bc6 allowed 15.Ng5 with mate in three.
Should I play ...h6 against Bg5 in the Modern Defense?
Timing matters. On move 9 the engine wanted 9...h6 (−0.66). On move 12, after I had castled, the same pawn move let White take on h6 with the bishop (+0.11 → +1.78).