Game #413: the Accelerated London System vs Bd6, and a 1191 who left after seven moves

On April 29, 2024, in game #413 of my rapid journey, I beat zafpar (1191) with White while rated 1008. The game lasted seven moves. After 1.d4 d5 2.Bf4, the Accelerated London System, Black answered with Bd6 to offer a bishop trade, I dropped my bishop back to g3, and after my seventh move Chess.com recorded the game as abandoned by my opponent. My rating went to 1020.

By rating gap, 171 points, it's tied with game #46 for the third-biggest upset in my first 2,116 rapid games. Game #46 also ended with my opponent abandoning. So two of the four biggest upsets in those 2,116 games were won by people leaving. I'm writing this one up anyway, because the seven moves are a clean answer to a question I kept getting wrong in the London: what to do when Black puts a bishop on d6.

Where I was that morning

This was the second game of April 29. Thirteen minutes earlier I had finished game #412, a messy win over krzysztofmielczarek (1144) where I left a mate in one on my back rank and my opponent missed it. The evening before, game #411 in the Accelerated London had gone the other way: a loss to bmw2228 (1008) with 57.6 accuracy, and Chess.com records that one as "game abandoned" too, with me as the player who didn't finish.

It was week 17 of the journey, the week I climbed back above 1000 and finished at 1017. My opponents that week were rated between 1005 and 1191, and zafpar was the 1191. It was also the highest rating I had beaten up to then: my best win before this one was against a 1173.

What the Accelerated London is, in one paragraph

The London System is White's set-up with pawns on d4 and e3, the dark-squared bishop out to f4, and the knight on f3. The "Accelerated" version just means the bishop comes out on move 2, before the knight. I played it a lot: 82 of my first 2,116 rapid games are filed under this name. Going into game #413, my record in it was 6 wins and 11 losses. It wasn't a weapon yet; it was a habit.

The moves were 1.d4 d5 2.Bf4 Nc6 3.e3 e6 4.Nf3. Black's knight on c6 sits in front of Black's own c-pawn. On move 4 the engine had my knight move and pushing the c-pawn to c4 tied for first, so the first four moves were all fine.

Moment 1: the Accelerated London System vs Bd6

Then Black played 4...Bd6, putting a bishop straight in front of mine.

r . b q k . n r
p p p . . p p p
. . n b p . . .
. . . p . . . .
. . . P . B . .
. . . . P N . .
P P P . . P P P
R N . Q K B . R

This is the position I want to remember. Black is offering to trade the dark-squared bishops. In the London, my bishop on f4 is the piece the whole set-up is built around, so I have three choices: take on d6, ignore it, or step back.

During the game analysis, the engine's preferred move was c3. When I checked the position again for this post, its top three were my move, bishop to g3 (+0.40), then a3 (+0.30) and c3 (+0.25), all scored as "about equal". So there was no wrong answer among the quiet ones. What I like about bishop to g3 is that it keeps the bishop on the same diagonal and makes Black decide. If Black takes, I get to choose how to recapture.

Moment 2: which pawn takes back

Black did take: 5...Bxg3.

r . b q k . n r
p p p . . p p p
. . n . p . . .
. . . p . . . .
. . . P . . . .
. . . . P N b .
P P P . . P P P
R N . Q K B . R

I took back with the h-pawn, the engine's first choice at +0.58. Taking with the f-pawn was rated -0.16, about equal. And the engine's third line is a warning for anyone who gets distracted: if I play bishop to b5 there without taking back, I'm simply a piece down, "black is completely winning" at -7.39.

The idea behind taking with the h-pawn is that my rook on h1 now stands on a half-open file pointing at Black's king's side, and my f-pawn stays home to protect e3. I'd be lying if I said I worked all that out in the game. I took towards the center because that's the rule I'd read, and the h-pawn happened to be the right one. I used 2 seconds on that move by the clock.

Moment 3: the pin, and then nothing

Black developed the other knight to f6, and I played my last move of the game.

r . b q k . . r
p p p . . p p p
. . n . p n . .
. . . p . . . .
. . . P . . . .
. . . . P N P .
P P P . . P P .
R N . Q K B . R

Bishop to b5 pins the knight on c6: the knight can't move without exposing Black's king behind it. The engine preferred c4 (+0.60) and scored my move at +0.34. Not an inaccuracy, just a slightly less useful move. After bishop to b5, the engine's best line for Black was to block with the bishop on d7, and it rated the position "about equal" (+0.41).

That was it. The game record ends with my seventh move. The game started at 09:04 UTC and Chess.com logged the end at 09:13. Between the two clocks, we had used about a minute and twenty seconds. I don't know why my opponent left, and the data can't tell me, so I won't guess.

Is a win by abandonment really an upset?

On paper, yes: Chess.com gave me 12 rating points for it, and my opponent was rated 171 points above me. In chess terms, no. The engine's evaluation never moved further than +0.62 from the start. Nobody was winning.

Abandoned games are not rare in my history. By the game records, 155 of my wins in those 2,116 rapid games ended this way. They count the same in the rating as any other win, but they don't teach me anything about converting a position. What this one does teach is the opening.

The lesson I took from game #413

When Black answers the Accelerated London System with Bd6, I don't need to panic about losing my favourite bishop:

1. Stepping back to g3 is fine. The engine rated it level with the quiet pawn moves a3 and c3.
2. If Black takes on g3, take back with the h-pawn. It opens the h-file for my rook and keeps my f-pawn guarding e3. In this position it was the engine's first choice by more than half a pawn.
3. Don't forget to take back. Any other move on move 6 just loses a piece.
4. Then play c4. The engine's first choice on move 7 was hitting the center, not pinning a knight.

What came next

The next game, #414, I lost with Black to bimalhembram (1035), with 52.4 accuracy. Then I won game #415 against julio_moscosokike2014 (1136). I finished the week at 1017, and I played the Accelerated London three more times that week. The game before this one is the back-rank lesson from game #412, the week is in the week 17 diary, and everything else is on the journey page. You can replay game #413 on Chess.com.

Seven-move games are easy to check by hand. Long ones aren't, and that's where Chess Coach helps: it finds the moments that mattered and explains them in plain words. Here's how the analysis works, and whether Claude can analyze chess games at all.