How to Review Your PGN Archive for Recurring Blunders (Not Just One Bad Game)

Reviewing a PGN archive for recurring blunders means reviewing each game twice: once yourself, once with an engine

Here's the direct answer. To find your recurring blunders in a PGN archive, review each game two times, not once. A blunder is a move that throws away a piece, or the whole game, in one shot. A PGN archive is just a folder of text files, one per game, in a format called PGN. PGN stands for Portable Game Notation. It's a plain text file that records every move you and your opponent made.

First pass: no engine. An engine, here, means a computer program that finds the best chess move in any position. WGM Dina Belenkaya, a Woman Grandmaster, recommends analyzing a game on your own before turning one on. Her reason is simple. You need to understand your own thought process during the game, before a machine tells you what you should have thought.

Second pass: bring in the engine. This is where you check your own read of the game against a computer's read of it.

Why two passes and not one? Strong players tend to follow the same loop, over and over. They play games. They review them carefully. They identify weaknesses. Then they train those weaknesses. Skip the review step. Or rush it. Either way, the loop breaks. You keep playing, but you stop learning from what you played.

That two-pass habit works fine on one game. Run it across a whole archive instead, maybe 30 games, maybe 100, and it starts to show you something. Something a single game never can. Which mistake you keep making, game after game. That's the real target of a PGN archive review. Not one blunder. The blunder that repeats.

Your own pass catches what an engine never sees: what you were actually thinking

An engine can tell you what the best move was. It can't tell you why you didn't play it. That's what your own pass is for.

As you replay a game from your PGN archive, slow down at the spots where you spent a lot of time. If you sat there for two minutes on one move, something was going on. Write down what you were thinking right at that point. Then write down which other moves you considered. Chess players call these candidate moves, the moves you looked at before picking one.

Do this for both sides of the board, not just your own moves. It's easy to only look at your own mistakes. But many positions hold a missed chance for both players, not just one. Seeing what your opponent missed, too, sharpens your read of the position.

By the end of this pass, you should have real notes. Not "I blundered on move 22." Instead, write down the real reason. What were you watching for at that moment? What did you miss, and why? That kind of note is gold later. It's the raw material an engine alone can never hand you, because an engine only sees the position on the board. It never sees what was going on in your head while you sat there looking at it.

Your engine pass should hunt for evaluation swings, not grade every move the same

Once your own notes are done, bring in the engine. But don't sit and grade every single move at the same level of attention. That wastes time.

Instead, scroll through the game fast, watching the evaluation number the engine shows. Bookmark any move where that number swings by 1.00 or more. That's a big swing, roughly the value of a minor piece changing hands in the engine's judgment. A swing that size almost always means something real happened, a missed tactic or a critical turn in the game.

At each bookmarked move, look at what grade the engine gives it. Most engines sort moves into a handful of labels, from Brilliant down to Blunder. Those labels are based on centipawn loss. Centipawn loss is how far your move was from the engine's own best move. It's measured in hundredths of a pawn, the tiny unit engines use to score a position.

Zoom out after you've bookmarked the swings. Check your average centipawn loss for the whole game, usually shortened to ACPL. ACPL is just the average of that centipawn-loss number across every move you made. A lower ACPL means your moves stayed closer to the engine's picks throughout the game. It's a rough signal, not a verdict. It's a fast way to tell whether a game was clean or shaky overall. Then you can dig into the specific swings that bookmarked it as worth reviewing.

One game tells you what happened; only a written log across games tells you what keeps happening

After each two-pass review, write one more thing down: the cause. Not just "I blundered my rook on move 30." Ask why, and pick from a short list of usual suspects. Did you not even consider the move that would have stopped it? Did you stop calculating too early, one move short of seeing the danger? Did you miss a forcing move, one that gives your opponent no real choice? Or did you just read the position wrong?

Write that cause down next to the game. That's your note for this one game. Do it every time you finish a review, and file it with that game's PGN.

Here's why that single note matters more once you have a whole archive. One game with a cause written next to it just tells you what happened in that game. It's a data point, nothing more. But go back through a batch of these notes, say two dozen of them, sitting in your archive. Read them side by side. Say "stopped calculating too early" shows up next to six different games. Those six games span six different openings, against six different opponents. That's not six unlucky moments. That's one habit.

You can't see that pattern from inside any single game, no matter how carefully you review it. You can only see it by laying your notes next to each other and counting. That's the real difference between reviewing a game and auditing an archive.

BlackBox Chess runs that same two-pass logic across your whole archive automatically

Everything above works. It also takes real time, game by game, note by note. If you'd rather skip the hand-logging across dozens of PGN files, that's exactly the gap BlackBox Chess fills.

BlackBox Chess is built for chess players under 2000 rating. It names your real recurring mistake. You don't get a folder of separate notes to compare by hand. Hand it your last 10 games. It runs Stockfish across all of them. Stockfish is a free chess engine, a computer program that finds the best move in any position, and it's one of the strongest engines around. It then tells you your number one recurring mistake, in plain terms. That's the free version. No account. No card. Nothing standing between you and the answer.

If that first answer is worth digging into, the paid version covers your last 100 games for $10. It doesn't just pile up more flagged moves. It ranks every mistake pattern by how much rating it's actually costing you, not just by how often it shows up. Each pattern comes with 3 real examples pulled straight from your own games, so you can see the exact positions, not a vague label. A fix course comes with it, built around whichever mistake ranks highest on your list.

Create a free account and you also get more. BlackBox emails you a daily drill. It's built from your own mistakes. That way, the work keeps going after the report closes.

The two-pass review above still teaches you something real about how to read your own games. But say you've got a folder of 50 PGN files and no free afternoon to work through them one at a time. The free 10-game check gets you the same kind of answer, without the hand-logging. Run it, and see your one recurring mistake named.