I came across this interesting position:

[Title "Black to move (with full castling rights)"]
[FEN "r2qkb1r/pp1nn1p1/7p/5N2/3p4/2P2N2/PP2QPPP/R1B3K1 b kq - 0 1"]

On first glance, it seems like Black ought to have a little bit of advantage, as White is down a rook for a pawn. But due to the significant danger for Black (under a pin and unable to castle safely), it is quite unclear, and even after looking at a couple of lines using Lichess analysis I am unable to decide which side is better off. Is there any broad-brushed strategy for either side that would gradually but decisively achieve victory?

Based on Stockfish evaluation, it seems that both sides can maintain a dead draw for 10+ moves, but I think the outcome under perfect play is unlikely to be a draw. So who is most likely winning here?

  • 1
    So others don't have to run it, Stockfish 11+ HCE from Lichess gives the position a -0.1 eval with the principal variation continuing 1...Nb6 2. Bf4 d3 3. Qe6 Nc8 4. Ne5 Qd5 5. Nd6+ Qxd6 6. Qf7+ Kd8 7. Rd1 Qf6 8. Rxd3+ Nd6 9. Qxf6 Jun 7, 2021 at 18:52
  • And SF 13+ NNUE from lichess likes that exact principal variation slightly more with a -0.5 eval at depth 35. Jun 7, 2021 at 23:30
  • 1
    I don't think what the engine says is too relevant here. Most humans would go 1...Nf6 instead. The maneouvre ...Nb6-c8 stopping the perpetual is almost out of reach for a non-master
    – David
    Jun 8, 2021 at 7:56

2 Answers 2


Is there any broad-brushed strategy for either side that would gradually but decisively achieve victory?

Sounds like what you are really asking here is "Can black survive long enough, without giving back too much material, to take advantage of the extra material to win." That's because white's winning strategy would have to be to checkmate black quickly and that's not a gradual process and in any case, as you say, the engines evaluate the position as close to equal out to move 10.

If there were such a strategy for black it would be something like:

  1. Avoid checkmate
  2. Avoid giving back too much material
  3. Get the queens off

1 is achievable, 2 is achievable for a while but white can easily sidestep 3.

Although a rook for a pawn up, black is facing an immediate checkmate threat, Nd6#, plus Qe6 looks really nasty, plus the white squares around the black king are very weak.

The most obvious move, Qb6, protects against both the checkmate and Qe6 but is one of many second favourites of the engine. As some of the comments point out, Nb6 is the first choice of the silicon-based players.

This carbon-based player likes the look of Nc5 which also protects against both the mate and Qe6 and aims to return some material to get to a more comfortable, albeit more materially balanced position, with a sequence like 1...Nc5 2. cxd4 Qd7 3. Nxe7 Bxe7 4. dxc5. Then black can choose between Rd8 threatening mate (albeit easy for white to parry) and O-O.

It could go something like this:

[FEN "r2qkb1r/pp1nn1p1/7p/5N2/3p4/2P2N2/PP2QPPP/R1B3K1 b kq - 0 1"]

1...Nc5 2. cxd4 Qd7 3. Nxe7 Bxe7 4. dxc5 Rd8 (4...O-O 5. Qc4+ Kh8 6. Be3 Bf6) 5. Be3 O-O 6. Rc1

Of course the engines don't go for any of this cowardly stuff. They prefer, as black, to hang on to the extra material as long as possible, even at the expense of having to make a number of "only moves":

[FEN "r2qkb1r/pp1nn1p1/7p/5N2/3p4/2P2N2/PP2QPPP/R1B3K1 b kq - 0 1"]

1...Nb6 2. Bf4 d3 {only move} 3. Qe6 Nc8 {only move unless you want to allow white a draw by repetition} (3...Qd5 4. Nd6+ Kd8 5. Nf7+ Ke8 6.Nd6+) 4. Ne5 Qd5 {only move} 5. Nd6+ Qxd6 {only move} 6. Qf7+ Kd8 7. Rd1 Qf6 8. Rxd3+ Nd6 {only move} 9. Qxf6 gxf6 10. Nf7+ Kd7 11. Nxh8 Nec8

At the end of the day the computer has given back material, black is up a piece for two pawns, but in return white has allowed the exchange of queens. White's more active pieces means that the evaluation in black's favour is less than 0.5.

I think the outcome under perfect play is unlikely to be a draw. So who is most likely winning here?

I think the outcome in such a double-edged position with human play is unlikely to be a draw. More scope for mistakes means humans are more likely to make them and lose, but the computer suggests that with perfect play it is a draw.

  • Interesting last statement. I actually believe that it should not be a draw even with perfect play, just that Stockfish is unable to see the correct answer. Of course, I have not a single bit of evidence to back that belief up.
    – user21820
    Jun 8, 2021 at 16:16
  • Hmm, maybe the evidence that it is a draw is that there are too many ways to reach a perpetual check. I guess I'll go with your answer then, thanks!
    – user21820
    Jun 9, 2021 at 16:00
  • A draw is possible if Black tries 1...Nf6 though
    – David
    Jun 9, 2021 at 16:07

This is a quite complicated position. It's not about "following a strategy" but rather about concrete calculation of different lines that could arise. Any "tiny" mistake by either side can lead to a huge change in evaluation (for example, White is threatening checkmate right now).

Most players would prefer to have White here, since playing with the initiative is often easier than finding defensive moves. For instance, after Black defends d6 with ...Nf6 or ...Nb6, White could follow with Bf4, insisting on Nd6+ and perhaps later Re1 to add pressure on the pinned knight. Qe6 is an option too with many chances for a perpetual (see for instance 1...Nf6 2.Qe6 Qd5 3.Nd6+). Finding natural moves for Black is not easy at all though.

  • Call me a materialist, but I'd actually prefer Black because Black is a rook up - White is the one who has to demonstrate compensation, while I know that as long as I don't get mated I will win.
    – Allure
    Jun 8, 2021 at 4:02
  • @Allure I'm afraid you'll hard a really hard time. What would you do to stop a plan like the one I described?
    – David
    Jun 8, 2021 at 7:55
  • You mean Bf4 intending Nd6+? I'd probably move the knight (...Nf6 or ...Nb6), then get ready to move the King after the knight check, or maybe even move it beforehand with ...Kf7 intending ...Kg8. Not pretty, but a rook is a rook. Checking with Stockfish, many of the candidate moves that first spring to mind would lose; on the other hand finding the refutation is not easy. What would you do after 1...Nf6 2. Bf4 Kf7?
    – Allure
    Jun 8, 2021 at 8:19
  • @Allure after 1...Nf6, White can get at least draw by repetition with 2. Qe6 followed by Nd6+. The only way for Black to fight for a win is 1...Nb6 with the idea or replying to 2.Qe6 with ...Nc8. What are the chances a human below 2200 ELO would find that plan in a live game?
    – David
    Jun 8, 2021 at 11:08
  • 1
    @Allure even in the lane you suggest, White can at least get a draw with 3.Nd6+ Kg8 4.Qe6+ Kh7 5.Ng5+
    – David
    Jun 8, 2021 at 11:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.