Most chess players analyze their games wrong. They plug a game into an engine, scroll through the moves, see where the evaluation changed, think "oh, I should have played that," and move on. Two weeks later, they make the exact same mistake.
Sound familiar? Here's how to analyze your games in a way that actually sticks — a process that turns raw engine output into real improvement.
If you're new to chess analysis, start with our guide on what centipawn loss means — it explains the numbers the engine throws at you. Then come back here for the full process.
The Problem with Engine-Only Analysis
Engines are incredible tools. Stockfish can see 30+ moves deep and evaluate positions with superhuman accuracy. But there's a fundamental problem: the engine doesn't know why you made a mistake.
When an engine shows you that 14.Nf5 was better than your 14.Be3, it doesn't tell you whether:
- You didn't consider Nf5 at all
- You considered it but rejected it for the wrong reason
- You were in time trouble and played too fast
- You misunderstood the position entirely
The reason for the mistake is what you need to fix. The engine only shows you the result.
This is why most players plateau despite analyzing every game. They treat analysis like checking answers on a test — looking at the right answer without understanding why they got it wrong. That's not analysis. That's scrolling. For a framework on what your accuracy score actually tells you at your rating level, see our accuracy benchmarks by rating guide.
A Better Analysis Process
Step 1: Replay Without an Engine First
Before opening any analysis tool, replay your game from memory. At each critical moment, write down (or mentally note):
- What was I thinking here?
- What alternatives did I consider?
- Where did I feel uncertain?
This takes discipline, but it's the most valuable step. Your own thought process during the game is information that no engine can provide.
A practical trick: after the game, jot down 3-5 moves where you felt uncertain. These are your "decision points" — the moments where analysis will have the highest return. Don't try to recall all 40 moves; focus on the moments of doubt. If you play on FireChess, the scanner flags these automatically by showing where your centipawn loss spiked, but even a quick mental note after the game helps.
Step 2: Identify Critical Moments
Not every move deserves deep analysis. Focus on:
- Turning points — where the evaluation shifted significantly
- Decisions — positions where you had 2-3 reasonable options
- Time pressure moves — moves you played quickly in a complex position
- Opening deviations — where you left known theory
Mark 5-8 critical moments per game. If you try to analyze every move deeply, you'll burn out and remember nothing.
Here's a real example of a critical moment from an Italian Game middlegame:
Italian Game middlegame — White's pieces (Ne5, Bc4, Qf3) all point at f7, but the critical move isn't a direct attack. The engine reveals d4! as best, opening the center for White's superior coordination. This is the kind of position where analysis catches moves you'd never find at the board.
White's knight on e5, bishop on c4, and queen on f3 all aim at f7. Your instinct might be to pile on with moves like Qg3 or Ng5. But analysis reveals that d4! is the critical move — it rips open the center, activates White's pieces, and leaves Black's passive Ba7 stranded. Without engine analysis, this quiet-looking pawn break is almost impossible to find over the board.
If you're struggling to identify which moments are actually critical, our guide on chess mistakes by rating shows the most common error patterns at every level — use those patterns to know what to look for.
Step 3: Use the Engine Surgically
Now turn on the engine, but only for your marked critical moments. For each one:
- Look at the engine's top move
- Understand why it's better (not just that it is)
- Compare it to what you played
- Identify the thinking error that led you astray
Example of good analysis:
"I played 14.Be3 because I wanted to develop my bishop. The engine preferred 14.Nf5 because it attacks g7 and d6 simultaneously, and Black can't defend both. My mistake was focusing on general development when there was a concrete tactical opportunity. I need to check for forcing moves before making quiet moves."
Example of bad analysis:
"Engine says 14.Nf5 is +1.2, I played 14.Be3 which is +0.4. Okay, noted."
The first analysis will change your thinking. The second won't.
The key distinction: good analysis identifies the cognitive error, not just the move error. "I missed a fork" is a move error. "I stopped calculating after my opponent's first reply because I assumed the position was simple" is a cognitive error — and it's the one you can actually fix. For a deeper dive on building this calculation habit, see our calculation training guide.
Step 4: Categorize Your Mistakes
After analyzing several games, patterns emerge. Common categories:
- Tactical oversights — missed forks, pins, skewers, discovered attacks
Here's a position where the tactical opportunity is right on the surface — but a rushed player walks right past it:
Black has just played …Be6, challenging White's bishop. Most players continue quietly with Be2 or Bb5, but analysis reveals the tactical shot Bxf7+! — after Rxf7 Ng5, White wins material with a devastating attack on the weakened kingside.
White can play Bxf7+! here — a temporary sacrifice that rips open Black's king. After Rxf7 Ng5, the knight attacks f7 and h7 simultaneously, and White's attack is crushing. A player who skips analysis sees a quiet position; the engine shows a forced winning sequence hiding in plain sight. This is exactly the kind of "I can't believe I missed that" moment that makes analysis worth doing.
- Positional misunderstandings — wrong plan, bad piece placement
- Calculation errors — saw the idea but miscalculated a variation
- Time management — good position but blundered under time pressure
- Opening preparation — played a suboptimal move in a known position
Knowing your dominant error type tells you what to study. If 60% of your mistakes are tactical, puzzle training will help more than studying strategy books. If most of your errors come from time trouble, fixing your chess time management is higher ROI than doing more tactics.
The FireChess scanner at /analyze breaks down your mistakes by category automatically — the "Accuracy by Phase" section shows whether your errors cluster in the opening, middlegame, or endgame, and the move badge distribution (how many Brilliant !!, Inaccuracy ?!, Blunder ??) tells you at a glance whether your problem is tactical blunders or gradual positional drift.
Step 5: Create an Action Item
Every analysis session should produce at least one concrete takeaway. Here's an endgame where the analysis takeaway is crystal clear:
White has a protected passed pawn on d5 and a more active king. The winning move is Kf5!, seizing control and winning the e5 pawn. Many players would passively wait here — analysis reveals the concrete path to victory.
White's protected passed pawn on d5 and active king give a decisive advantage. The winning move is Kf5! — taking the e5 pawn and using the passed d-pawn to promote. Many club players would play passively here, shuffling the king without a plan. Analysis reveals the concrete winning technique: activate your king, attack the weak pawns, and push your passer. This kind of endgame position is easy to draw if you don't know the method — and easy to win once analysis shows you the plan. If endgames are where you struggle, our endgame patterns guide covers the most common positions club players get wrong.
Your action items should be specific and behavioral, not vague:
- "Before playing a quiet developing move, check all captures and checks first"
- "In IQP positions, always consider d4-d5 pawn breaks"
- "Stop playing a4 in the Sicilian — it weakens b4 and doesn't achieve anything"
Not this: "Play more carefully" or "don't blunder." Those aren't actionable — they're wishes. A good action item changes what you do at the board, not what you wish you'd done.
Write it down. Review your action items before your next game. This is how analysis translates into better play.
Common Analysis Mistakes (Yes, You Can Analyze Wrong)
Even players who analyze regularly often fall into habits that waste time or reinforce bad thinking. Here are the most common ones:
Confirmation bias. You think your move was right, so you look for engine lines that support it and ignore the ones that don't. This is especially dangerous in positions where the engine gives two moves with similar evaluations — you pick the one that matches your original choice and call it "confirmed." Break this habit by always looking at the engine's top choice first, before checking what you played.
Stopping after the first engine line. The engine shows +1.5 after Nf5. You note it and move on. But you never asked: what does Black play after Nf5? What's the best defense? If you don't understand the opponent's resources, you'll play the "engine move" at the board and be surprised when your opponent finds the defense you never looked at. Always play out 3-4 moves of the opponent's best response.
Analyzing only the moves, never the clock. A 15-second blunder and a 5-minute blunder have different root causes. The first is usually a hand-slip or mouse error; the second is a thinking-process failure. If you don't look at when you spent time, you can't fix why you went wrong. Most online platforms show move times — use them.
Ignoring draws and wins. You blundered a piece in a loss, so you analyze that game. But you also missed a mate-in-2 in a win and drew a game you were winning by +5. Those are equally instructive. The accuracy score captures this — a win with 65% accuracy tells a very different story than a win with 95% accuracy.
No follow-up. You write "stop hanging pieces on move 12" and then play 20 more games without ever addressing it. Analysis without follow-through is entertainment, not training. Your action items should feed directly into your next puzzle session, opening review, or endgame practice.
How Many Games Should You Analyze?
Quality over quantity, always.
Deeply analyze 2-3 games per week rather than superficially checking 20. A thorough analysis of a single game — where you understand every critical moment — is worth more than a quick engine check of ten games.
For players who want to improve consistently:
| Games Played per Week | Games to Analyze Deeply | Quick Engine Scan |
|---|---|---|
| 3-5 | 1-2 | All |
| 5-10 | 2-3 | All |
| 10-20 | 3-4 | Losses + interesting wins |
| 20+ | 4-5 | Losses only |
Always prioritize your losses and unexpected draws. Wins feel good but teach you less.
Bulk Analysis: When It Makes Sense
Deep, manual analysis is ideal but time-consuming. There's a complementary approach: bulk scanning across many games to find patterns.
Instead of analyzing each game individually, you scan 50-100 games and look for:
- Repeated positions where you consistently err
- Average accuracy by game phase (opening, middlegame, endgame)
- Tactical pattern gaps — specific motifs you keep missing
- Rating correlation — whether your losses have common themes
This bird's-eye view reveals systemic weaknesses that game-by-game analysis can miss. You might discover that your accuracy drops sharply after move 35, suggesting endgame study is needed. Or that you consistently lose 40 centipawns in certain opening structures. Our guide on finding opening weaknesses shows how to use this pattern data to target your opening study.
If you want to estimate your opponent's rating from the PGN before running the engine, our Guess Elo from PGN guide shows you how to read the signals manually. FireChess's scanner at /analyze is built for this — upload your PGN and it shows repeated mistakes, opening leaks, and accuracy trends across your last 50-200 games in one view. The "Opening Leaks" section groups every position you've played more than once and shows your average centipawn loss in each, so you can see exactly which opening lines are costing you rating points.
The Best Tools for Analysis
For Deep Analysis
- Lichess Analysis Board — free, uses Stockfish, great for move-by-move review
- Chess.com Game Review — provides human-readable insights (premium feature)
- A physical board — seriously, replaying games on a real board improves memory
For Bulk Pattern Detection
- FireChess — scans your games for repeated mistakes, opening leaks, missed tactics, and endgame errors. Runs Stockfish 18 in your browser for privacy.
- Chess.com Insights — shows trends over time (premium feature)
- Custom PGN analysis — for tech-savvy players who want to build their own scripts
For Targeted Training
- Puzzle training (Lichess, Chess.com) — fixes tactical weakness
- Endgame trainers (Lichess Practice, chess endgames app) — fixes endgame weakness
- Opening trainers (Chessable, drill mode tools) — fixes opening weakness
If you're deciding which platform to use, our Lichess vs Chess.com improvement comparison breaks down the analysis features of each. And if you want to build a complete study system around your analysis, our chess study plan guide shows how to turn analysis findings into a structured training routine.
Putting It All Together
Here's a weekly analysis routine that works:
Monday: Quick scan — review weekend games with an engine, note big mistakes. Use FireChess's scanner to flag the high-centipawn-loss moves instantly.
Wednesday: Deep analysis — pick your most instructive loss and go through the 5-step process. Budget 20-30 minutes. Write your action items.
Friday: Pattern review — review your action items from the week, scan your recent games for recurring patterns. Check if the same mistake categories keep appearing.
Weekend: Play and apply what you've learned. Before each game, glance at your action items — even 30 seconds of reminder changes how you play.
The key is consistency over intensity. Thirty minutes of focused analysis three times a week beats a five-hour marathon once a month.
Frequently Asked Questions
Q: How long should I spend analyzing a single game?
For a thorough deep analysis following the 5-step process, budget 20-30 minutes per game. That breaks down to roughly 5 minutes replaying without an engine, 5 minutes identifying critical moments, 10-15 minutes on engine-assisted analysis of those moments, and 5 minutes writing your action items. If you're short on time, even a focused 10-minute review of your 3 biggest mistakes is far better than nothing. The goal isn't to analyze every move — it's to extract lessons that stick.
Q: Should I analyze my wins or just my losses?
Always prioritize losses and unexpected draws, but don't ignore wins entirely. Wins where you were winning from the opening teach you little. However, wins where you were worse and came back, or where you missed a faster finish, are highly instructive. A good rule of thumb: analyze 70% losses, 20% hard-fought wins, and 10% draws. You can often learn more from a loss where you had a good position and blundered than from a clean win where your opponent blundered on move 8.
Q: What's better: engine analysis or analyzing without an engine first?
Neither alone — they serve different purposes. Analyzing without an engine first (Step 1) captures your thought process and reveals why you made decisions. Engine analysis (Step 3) shows you what the objective best move was. Doing only engine analysis means you see the right move but never understand why you missed it. Doing only manual analysis means you might reinforce incorrect evaluations. The 5-step process combines both: think first, then check with the engine, then reconcile the difference.
Q: How many games do I need to analyze before I see improvement?
Most players notice a difference after analyzing 10-15 games deeply — typically 4-6 weeks of consistent work. But the real improvement comes from acting on your action items. If your analysis reveals you consistently miss tactical shots on the back rank, and you train back-rank puzzles for two weeks, you'll see results quickly. If you analyze 50 games but never change your training or habits, you won't improve. Analysis without follow-through is just entertainment.
Q: Can I use AI or automated analysis instead of doing it manually?
Automated tools like FireChess are excellent for bulk scanning — spotting patterns across dozens of games that you'd never catch manually. But they can't replace the self-reflection step. An AI can tell you that you consistently miscalculate in time trouble, but only you can decide whether that's because you panic, spend too long on earlier moves, or don't know the positions well enough to play quickly. Use automated tools to identify what your problems are, then use manual analysis to understand why they happen and build a fix.
Q: What is the best free tool for chess game analysis?
Lichess offers the best free analysis experience — unlimited Stockfish analysis, no paywalls, and a clean interface. For bulk pattern scanning across many games, FireChess's scanner at /analyze is free and runs entirely in your browser. Chess.com's Game Review is useful but most features require a premium subscription. The best approach is to combine tools: use Lichess for deep single-game analysis and FireChess for spotting patterns across your entire game history.
Q: How do I stop making the same mistakes after analyzing?
The gap between analysis and improvement is application. After identifying a pattern — say, you consistently miss back-rank mates — you need to train that specific pattern. Do 20-30 back-rank puzzles on Lichess or Chess.com. Then, before your next game, remind yourself: "check for back-rank threats." The combination of analysis (identifying the problem), targeted training (building the pattern), and pre-game reminders (activating the pattern) is what actually breaks the cycle. Analysis alone just tells you what's wrong — it doesn't fix it.
Start Analyzing Smarter
If you want to jumpstart your analysis, try scanning your last 50 games with FireChess at /analyze. In a few minutes you'll see your repeated mistakes, accuracy trends, and weakest areas — giving you a clear roadmap for improvement. The basic scan is free, and everything runs privately in your browser.