How to Analyze Your Own Shogi Games

Intermediate · Improvement

The single highest-return habit for improving players — a repeatable method to learn from every game you play.

Once you know the rules, openings, and basic tactics, the fastest way to get stronger is not playing more games — it is reviewing the games you already played. Strong amateurs and pros all do it. The good news: you don’t need a teacher. You need your game record and about ten focused minutes per game.

This guide gives you a concrete, repeatable review method. Work through it with one of your own games open in the interactive board further down the page.

Table of contents

What You Need: Your Game Record (Kifu)
The 5-Step Review Method
Replay Your Game Here
A Worked Example: “Castle First”
Turn Analysis Into a Study Loop
Keep Going

What You Need: Your Game Record (Kifu)

A kifu (棋譜) is the move-by-move record of a game. Most apps and servers let you export or copy it. You will see it in one of three formats:

KIF / KI2 — the Japanese notation, e.g. ▲7六歩 △3四歩 ▲2六歩 or 　1 ７六歩(77).
USI — the engine notation used by many apps, e.g. 7g7f 3c3d 2g2f.

All three work in the board below — just paste and replay. If your game is only on a screen, you can also re-enter the moves by hand.

The 5-Step Review Method

Replay the whole game once, quickly

Don’t analyze yet. Just step through it and recall the story: when did it feel comfortable, and when did it start to slip? Your feelings during the game are the first clue to where the real mistakes are.

Find the turning points

A game is usually decided by two or three moves, not forty. Mark the moments where the advantage clearly changed hands — a piece you lost for nothing, an attack that fizzled, a defense that came one move too late. Those are the moves worth studying.

Ask three questions at each turning point

Pause on the move and answer honestly:

What did I miss — a threat, a better move, or a whole plan?
Was it a one-move oversight, or a slow drift over several moves?
What is the rule or pattern I can take to the next game?

Classify the mistake

At the improving level, most losses fall into four buckets. Naming the bucket tells you what to study next:

Opening / preparation — you left the book early or your shape was bad. → Study joseki.
King safety & timing — you attacked before your castle was finished. → “Castle first” (玉の堅さ).
Reading (yomi) error — you simply miscalculated a sequence. → Tsume and tactics.
Endgame speed — you grabbed material when you needed tempo. → “In the endgame, speed beats material” (終盤は速度).

Let an engine check your reading — the right way

An engine is a fact-checker, not a coach. Use it to confirm whether your turning point really was the mistake, then go back and understand why in human terms. Read the evaluation as a win probability, not a centipawn count: a swing from “even” to “clearly worse” is a real mistake; a 30-point wobble in an already-won position is noise. Don’t try to memorize the engine’s every move — extract the one idea you missed.

Replay Your Game Here

Paste your kifu (KIF, KI2, or USI) and step through it. Use the move list to jump straight to the turning points you marked in Step 2.

Tip: if nothing happens, check that the game is an even game (hirate) — handicap games aren’t supported yet.

A Worked Example: “Castle First”

The most common turning point in improving-level games is an attack launched before the king is safe. The proverb is blunt: build your castle before you start fighting. When you review, look specifically for the move where you (or your opponent) opened hostilities with the king still sitting in the center.

Two classic proverbs to keep in mind as you review:

開戦は歩の突き捨てから — “Open the battle with a pawn sacrifice.” Before a real attack, you usually need to open a file or a diagonal. If your attack stalled, check whether you skipped this preparation.

終盤は駒の損得より速度 — “In the endgame, speed beats material.” Late in the game, being one move faster to the enemy king is worth more than a captured piece. Many “I was ahead and still lost” games are a defender’s tempo lost here.

Turn Analysis Into a Study Loop

The point of review isn’t to feel bad about one game — it’s to build a short list of recurring mistakes. Keep a running note like this:

Pattern: “I attack before castling.” → Fix: finish the Mino/Yagura first.
Pattern: “I miss back-rank mates.” → Fix: 5 tsume puzzles a day.
Pattern: “I take a pawn in the endgame instead of attacking.” → Fix: count the race, not the material.

After ten reviewed games, your list will repeat itself. Those repeats are exactly what to train — and they are the difference between staying at your rank and climbing out of it.