Should I analyze my chess games with or without Stockfish?
Should you analyze your chess games with or without Stockfish? Both, and in that order. Spend about five minutes on your own first, writing down what you were thinking, then turn the engine on. The two methods catch different mistakes: in my own 2,116 rapid games on Chess.com, 43% of my blunders were not punished by my opponent's next move, so I would never have found them without an engine; and no engine can tell me why I chose a bad move.
This guide explains what each way of analyzing catches, what it misses, and a short routine that uses both. The examples are real positions from my account (nacsol3), checked with Stockfish, from the months when I was stuck around 1000.
A two-pass worksheet for your next finished game
Use this as a review exercise, rather than a rule about how many minutes every player needs. Before opening the engine, choose one moment you found difficult and write three short notes: what you expected, the alternative you considered, and what reply worried you. Leave room for the computer check beside each note.
Then open Stockfish and test your actual candidate, including the opponent's strongest reply. If your move was sound, keep the reason; if it failed, record the first reply that breaks it. Finally hide the line and try the position again. A note such as “I forgot that the pawn can capture my knight” is easier to use in a future game than a copied evaluation number.
This follows the order in Dina Belenkaya's game-analysis guide, checked October 31, 2026: your own review first, computer verification afterwards. It does not mean your first conclusion must be right. The useful comparison is between your written thought and the verified continuation. If you cannot explain the first move of that continuation, return to the board and ask what it attacks or prevents before studying the rest.
What each method catches (and misses)
| Without an engine | With Stockfish only | Both, in order | |
|---|---|---|---|
| Finds why you chose a move | Yes, if you write it down | No | Yes |
| Finds tactics you never saw | Rarely: you miss them again | Yes | Yes |
| Finds blunders your opponent didn't punish | Almost never | Yes | Yes |
| Spots "good-looking" moves that lose | No | Yes | Yes |
| Trains your own calculation | Yes | Little | Yes |
| Time for a 10-minute game | 5–15 min | 5 min | 15–20 min |
| Main risk | Wrong conclusions | Copying moves without learning | Takes a bit longer |
The short version: going through the game on your own gives you the story of the game, and Stockfish (the free, open-source engine behind Lichess and most analysis boards) gives you the facts. You need the story to learn something you can reuse, and the facts to make sure the story is true.
What analyzing without an engine gives you
When you replay a game on your own, you remember things no engine can see: that you were short of time, that you were scared of a check, that you thought a trade was forced. That is the part you can actually change next game.
Coaches agree on this order. In Chess.com's guide to game analysis, WGM Dina Belenkaya makes engine-free review the first step and puts the engine last, as a "sparring partner" to test your conclusions (How To Analyze Your Games, Chess.com, July 10, 2025). Her step one fits in a line: "Always analyze your game on your own first."
What it can't do
The honest problem is that you analyze with the same eyes that played the game. A Chess.com forum post makes the point well: the author had been told to review without an engine, and asked what he could possibly see afterwards that he missed at the board, since his puzzle results showed how much he missed. Eighteen months later he wrote that he still rarely found what the engine showed him (Chess.com forum, "Non-Engine Self Analysis vs Puzzle Performance?", September 12, 2022).
My data says the same. In my games, when I blundered, the opponent's very next move failed to punish it 2,012 times out of 4,723 (I count a reply as "not punishing" when it wasn't the engine's best move and gave back at least 5 points of winning chance, the same line Chess Coach uses for an inaccuracy). In those games nothing visible happened: no piece was lost, the game went on. Replaying them on your own, you have no reason to stop there.
What Stockfish gives you
The engine is fast and doesn't get tired. It shows every big swing in the evaluation graph, labels each move (inaccuracy, mistake, blunder; here's what those labels mean) and gives you a better move with a line behind it. On Lichess this is free: its analysis board runs Stockfish on any game. Chess.com's Game Review is limited on free accounts (how many free reviews you get).
What it can't do
Stockfish tells you what was wrong, never why you chose it. A line like "Nb5 Nxf3+ Qxf3 Qe5 Bd3 Nf6, +4.56" is correct, but you can't take a list of moves into your next game. You can take a sentence like "my rook move left my back rank empty". Getting from the first to the second is your job. If you need help reading the numbers themselves, see how to read Stockfish analysis.
The second risk is skipping your own thinking. If you open the engine first, the green arrow appears before you've tried to find anything, and it is very hard to "unsee". The Chess.com article and the forum post above both make this point.
Example 1: the move I would have called brilliant
March 1, 2024, game #244 of my rapid series. I'm White against Anixamander (1073) in a 20-minute game (game on Chess.com). Black's queen has just taken my pawn on c2. I'm clearly winning.
What I played (red arrow): my rook went to c1 (Rc1). It attacks the queen and, behind it, the rook on c8: a double attack, the kind of move you feel proud of.
What Stockfish says: it's a blunder. Before it I was at +6.41, completely winning; after it, −4.73. The engine's line explains why: the queen takes my rook on c1, I take back with the other rook, and Black's rook takes on c1 with check. My king has no escape square behind its pawns (a "back-rank" problem), so my best answer is to block with the bishop, and Stockfish rates the result as winning for Black.
The better move (green arrow): rook to d7 (Rd7), attacking the bishop on b7. Stockfish's line wins that bishop and keeps a big advantage.
Why this matters for the question: my opponent didn't find the refutation either. They took a pawn on b3 instead, and the game went on until move 68, which I lost. If I had reviewed this game only on my own, I would have marked Rc1 as one of my best moves. Without the engine, a losing move that "worked" looks exactly like a good one.
Example 2: the chance I never knew I had
March 4, 2024, game #253. I'm White against DannoG73 (997), 20-minute game (game on Chess.com). Black's knight has just taken my knight on e5. I have two ways to take back.
What I played (red arrow): bishop takes the knight (Bxe5). It felt natural: recapture, keep the material level.
What Stockfish says: the pawn recapture (dxe5) wins a piece. The pawn takes the knight and attacks two black pieces at once, the bishop on d6 and the knight on f6. That's a fork: a double attack where your opponent can only save one. Stockfish rates dxe5 at +5.82; my move left the game about equal (+0.09). In winning chances, I went from 90% to 49% in one move.
Why this matters for the question: I had more than 18 minutes on my clock and nothing in the game told me anything had happened. A missed win doesn't leave a mark: no piece hangs, no alarm goes off. On my own I would have replayed 7.Bxe5 without stopping. This is the kind of moment that only an engine finds, and once found, it makes a good puzzle to repeat (puzzles from your own games).
Should I analyze my chess games with or without Stockfish? A routine that uses both
Here is the order I'd suggest after every rated game. It takes 15 to 20 minutes for a 10-minute game; the full version, step by step, is in how to analyze chess games as a beginner.
- Without the engine (5 minutes). Replay the game. Write down two or three moments where you were unsure, what you were trying to do, and where you think the game turned.
- Turn Stockfish on and open the evaluation graph. Find the two or three biggest jumps. Compare them with your notes: did you notice those moments?
- For each jump, cover the answer. Look at the position and try to find the better move yourself before you look at the arrow.
- Say what you missed in one plain sentence. "I didn't check what my opponent could capture with check" is useful. "Stockfish preferred Rd7" is not.
- Look for the moments you didn't feel. Scan for your opponent's mistakes that you didn't punish, like Example 2. Those are invisible without the engine.
The comparison in step 2 is the most useful part. The moments in your notes and in the graph are mistakes you can already feel; you need practice. The ones in the graph but not in your notes are blind spots; you need new patterns. The ones in your notes but not in the graph were fine; you can stop worrying about them.
When to lean more one way
- No time at all? Use the engine and look only at the biggest swing. A two-minute check beats no review.
- A long game (30 minutes or more), or over the board? Give your own review more time. You thought more during the game, so your notes are worth more.
- Same mistake again and again? You don't need more analysis; you need practice on that pattern. Turn the positions into puzzles.
- Learning an opening? Your own review won't tell you what the main line is. Use the engine or an opening explorer, and check where you left theory.
Common mistakes
- Opening the engine first. Once you've seen the arrow, you can't test yourself any more.
- Reading every line for every move. Most small inaccuracies at club level don't decide the game. Spend your time on the two or three biggest swings.
- Only reviewing losses. Example 1 was a winning position and Example 2 looked like a quiet move. Wins and "normal" games hide blunders too.
- Trusting your own review as the final word. Your conclusions can be wrong. Check them, especially the ones you're proud of.
- Copying the engine's move without the reason. If you can't explain the green arrow in one sentence, you haven't learned it yet.
How Chess Coach helps
Chess Coach is a connector for Claude and ChatGPT that reviews your finished Chess.com or Lichess games with Stockfish on our server. It marks the key moments ①②③ on a board with arrows (red for the move played, green for the better one), shows the evaluation graph, and explains each moment in plain words. You can try your own moves, press "Why?" to ask your assistant, and use Practice mistakes mode to retry the positions you got wrong. It fits step 2 onwards of the routine: do your five minutes alone, then ask for the review. The free plan includes one in-depth review per day; see how to connect it or the AI chess analysis page. It never helps with games in progress.
Frequently asked questions
Should I analyze my chess games with or without Stockfish?
Both, in that order. Spend about five minutes going through the game on your own and writing down what you were thinking, then turn Stockfish on to check your notes and find the moments you didn't notice. Each method catches mistakes the other one misses.
Is it bad to analyze chess games with an engine?
No. The risk is only using the engine: you see the right move, but not why you missed it, so the same mistake comes back. Using Stockfish after your own review keeps the engine's accuracy and adds your own reasoning.
Can I analyze my chess games without an engine at all?
You can, and it trains your calculation, but at club level you will miss a lot. In my games, about 4 in 10 of my blunders were not punished by the opponent's next move, so the result never told me they happened. An engine is the only quick way to find those.
How long should I spend on each part?
For a 10-minute game, about 5 minutes without the engine and 10 minutes with it is enough. Focus the engine time on the two or three biggest swings in the evaluation graph instead of every move.
Do I need to pay to analyze with Stockfish?
No. Lichess runs Stockfish analysis for free, and Chess.com gives a limited number of free Game Reviews. Chess Coach also uses Stockfish, on our server, and its free plan includes one in-depth review per day.