How to Analyze a Chess Position Online: FEN, Evaluations and Engine Lines

Sometimes you don't want to review a whole game — you want to know about one position. A puzzle you couldn't solve, a moment in a book, a position a friend sent you. The fastest way to hand a single position to an engine is a FEN, a one-line text description of the board.

A FEN has six parts separated by spaces. The first describes the pieces, rank by rank from the eighth down to the first, using letters for pieces — uppercase for White, lowercase for Black — and numbers for runs of empty squares. The second is whose turn it is: w or b. The third lists castling rights, such as KQkq, or a dash if nobody can castle. The fourth is an en-passant target square, if there is one. The last two are counters for the fifty-move rule and the move number.

The two parts people most often get wrong are the side to move and castling rights, and both change the engine's answer completely. If you set up a position where White is to move but the FEN says b, the engine will analyse the position for Black. If you forget to remove castling rights after the king has moved, it may suggest castling in a position where it is illegal.

You rarely need to write a FEN by hand. Any chess site can export one from a game or a board editor, and ChessDeeper's position analyzer accepts a pasted FEN directly. The main reason to understand the format is to check it quickly when an engine's answer seems strange.

Once the position is loaded, the engine reports an evaluation in pawns: +1.00 means White is better by about a pawn, negative numbers favour Black, and a value shown as M3 or #3 is a forced mate in three. Treat anything within about three tenths of zero as equal. The exact decimals matter far less than the direction and the size of the change between moves.

Watch the depth as well as the number. An engine searches progressively deeper, and its evaluation after a fraction of a second can be very different from its view a few seconds later, particularly in positions full of tactics. Let it run until the best move stops changing before you trust it.

The most useful thing an engine shows is not the number but the line — the sequence of moves it expects. Play through it. If the engine says a move wins, the line tells you why: which threat it creates, what the defence is, where the material falls. Reading the line is how an engine teaches you something; staring at the evaluation is how it merely tells you.

Finally, compare more than one candidate. Looking at the second-best move alongside the best one tells you whether the position had one narrow path or several reasonable ones. A position where only one move holds is a genuinely hard position, and missing that move is a very different mistake from missing a small improvement in a comfortable position.