To review chess games with Stockfish, set depth 22-25 in Lichess Analysis Board, focus on moves marked "?" or "??", identify critical moments where evaluation shifts more than 1.5 centipawns, log 3 key error patterns per game, and design corrective drills for the next week. One reviewed game outweighs ten unreviewed games.
Setup Stockfish on Lichess and other online chess platforms
Lichess: open the game in the analysis board, click the "Engine" toggle, choose depth 22-25 (the default 18 is too shallow for amateur review). On Mac and PC desktop, browser-based Stockfish runs to depth 30 plus at reasonable speed.
Online chess platforms: use the "Game Review" feature (Premium) or the "Analysis" tab. Depth settings sit behind the gear icon. A Premium account unlocks deeper cloud Stockfish plus natural-language explanations per move.
For serious review (post-tournament, long Rapid games), consider installing Stockfish 16 locally with a GUI like Cute Chess or Arena. A local install runs depth 30 plus with multi-core CPU.
Reading the Evaluation: Centipawn, Mate, Best Move
Stockfish evaluation comes in centipawns (cp). +1.0 means White is ahead by one pawn (or equivalent advantage). +3.0 plus usually signals a technically winning position. Mate-in-N means forced mate in N moves.
Stockfish’s best move is sometimes unrealistic for amateurs (deep 7-move tactical, or positional moves whose payoff only appears 15 moves later). Avoid obsessing over the best move; focus on moves that are reasonable by principle and avoid moves that drop evaluation by more than 1.0.
Centipawn loss (CP loss) per move is a useful aggregate metric. Master-level players average under 30 CP loss per game. A 1500 amateur averages 60-80. Improvement target: drop 10-20 CP loss every 3 months.
Identifying Critical Moments
A critical moment is a move where evaluation shifts significantly (more than 1.5 centipawns). These usually mark a tactical blunder or a strategic mistake that changed the balance.
Focus on the 3 biggest critical moments per game. Ask yourself: "What was I thinking at that point? Which pattern did I miss?" Write the answer in your review journal (physical notebook or Notion).
Critical moments often appear at transitions (opening to middlegame around moves 12-16, middlegame to endgame around moves 28-35). These phases shift the strategic plan and trip up many players.
Logging Error Patterns: A Simple, Useful Table
A simple journal layout: Date, Opponent Rating, Opening, Critical Moment Move, Pattern Error (e.g., "missed back-rank mate", "traded pieces without a goal", "wrong rotation timing"), and Lesson to remember.
After 20 reviewed games, scan for repeated patterns. If "missed back-rank mate" shows up 8 times, that is a strong signal for a 2-week themed puzzle drill on back-rank patterns.
A consistent review journal carries more weight than the engine analysis itself. The engine delivers data; the journal delivers reflection. Without reflection, learning never transfers to the next game.
Common Mistake: Engine Worship and How to Avoid It
Engine worship is blind faith in Stockfish’s "best move" while ignoring that the engine has a horizon (depth limit) and a style (computer-like, often counter-intuitive). Some engine best moves require 15-move-deep calculation that is beyond human reach.
Focus on the "human best move" instead of the "engine best move". Ask: "Could I find this move with 10 minutes on the clock? What pattern should I train to spot this next time?"
Resist reviewing only losses. Reviewing wins matters too, so you can identify lucky wins (where the opponent blundered, while you also played poorly). Without reviewing wins, you stay over-confident and stuck.
Frequently Asked Questions
Can an engine replace a human coach?
An engine excels at flagging technical errors (tactics, position evaluation). A human coach excels at diagnosing thinking-pattern errors, mental game, and providing a personal roadmap. Combining both is optimal: engine for material, coach for meta-cognition.
Which Stockfish depth is enough for amateur review?
Depth 22-25 suffices for amateurs (up to 2000). Deeper (30 plus) helps in deep endgame study or complex positions. For daily review, 22-25 balances accuracy and speed.
How do I review fast blitz games?
Blitz review should focus on opening accuracy (moves 1-12) and critical blunders (dropped piece, missed basic tactic). Avoid scrutinising every blitz move because thinking time was short and many moves are intuition-driven and hard to analyse.
Do I need the latest Stockfish 16?
Optional. Stockfish 15 or Lichess cloud Stockfish already outclasses any amateur. The latest version matters for top GMs and engine matches. For amateur review, the free version on Lichess is more than enough.
How long should one game review take?
20-30 minutes for a serious Rapid game. Skim the opening (5 minutes), focus on middlegame critical moments (15-20 minutes), quick scan the endgame (5 minutes). Longer slips into diminishing returns because you over-analyse insignificant moves.