Poker hand history exports 15 fields — the read needs 40

Poker hand history exports are standardised to roughly 15 fields, and that number has barely moved in two decades. The problem is that a competent read of a single decision needs closer to 40. Everything in between — timing, bet-sizing history, table image, the thing your gut noticed on the turn — falls outside the export format, which means the hand you review is never quite the hand you played.

This isn't a complaint about software. It's a structural gap, and it has a specific cost: the gap between what you remember and what the file says is where most players' study time quietly leaks away.

What the 15 fields actually capture

The PokerStars hand history format, which most other rooms either license, clone or approximate, gives you a tight bundle per hand. You get the hand ID and timestamp, the table name and stakes, the button position, each player's starting stack, the hole cards (yours, and opponents' only at showdown), the blinds and antes, and then the action in sequence: who folded, who called, who raised, and to what. You get the board, the pot size at each street, the result, and the rake. Add the showdown reveal and the winner, and you're at the 15 or so fields that have been the industry default since the mid-2000s.

That's a ledger. It's excellent at answering "what happened" and "what did it cost." It is close to useless at answering "why did I do that," which is the only question that matters when you're reviewing a losing session at 11pm on a Tuesday.

The gap is not a secret. Every tracker — Holdem Manager, PokerTracker, DriveHUD, the newer web-based ones — builds its own database on top of the raw file and infers the missing 25 fields from the action. Inferred, not recorded. That distinction matters more than most players realise.

The 25 fields the file never sees

Timing and tempo

A hand history records that a player raised to 3.5bb on the turn. It does not record that they took 14 seconds to do it, or that their average decision time in the session was 2.8 seconds, or that this was the first time in 90 hands they'd tanked. Timing tells are the single most reliable live read and among the most reliable online ones, and the export format has no field for them. Trackers approximate with a "time to act" column if the room's client logs it, but coverage is patchy and inconsistent across skins.

Bet-sizing history within the session

You know a player's VPIP and PFR from the HUD. You don't know from the export that they've used a 33% continuation bet on every dry flop for the last hour and then suddenly fired 80% on a wet one. The file shows the size; it doesn't show the deviation from that player's own baseline, because the file has no concept of a baseline. It's a series of isolated snapshots pretending to be a narrative.

Stack dynamics and table flow

The export tells you the stack at the start of the hand. It doesn't tell you that the player to your left just doubled up and is now playing 40bb behind with a short-stack mindset, or that the table broke from six-handed to heads-up three hands ago. Table composition changes are invisible. So is the fact that you'd been card-dead for 40 minutes and your image was, accurately, "hasn't played a hand."

The action you didn't take

This is the one that stings. A hand history records the bets that happened. It cannot record the bet you considered and rejected, the three-bet you almost made, the fold you talked yourself out of. Your own decision process — the actual thing you're trying to train — leaves no trace. Reviewing a hand history is like reviewing a chess game where only the moves played are recorded and the clock, the player's mood, and the three candidate moves they rejected are all missing.

Why the format hasn't moved

Partly it's inertia: the format is a de facto standard, and changing it breaks every parser, every tracker and every back-office reconciliation tool in the ecosystem. Partly it's that rooms have no commercial incentive to hand you richer data — timing tells, in particular, are information the room holds and you don't, and there's no obvious reason for them to surrender it. And partly it's that the 15 fields were designed for dispute resolution and rake calculation, not for study. They do their original job perfectly. They were never asked to do this one.

The workaround most serious players land on is manual annotation: a notes field, a spreadsheet column, a voice memo after the session. It works, in the way that keeping a food diary works — effective, tedious, and abandoned by roughly everyone within a fortnight. The 2023 proliferation of solver-integrated trainers has made this worse, not better, because the trainer will happily grade your river bet against a GTO baseline while remaining completely blind to the fact that you'd been drinking since the second break.

What the gap costs you

Here's the numerical anchor worth sitting with: a player grinding 25,000 hands a month at a 6-max table sees roughly 4,000 hands go to a flop. If they review 20 of those — a generous figure for most — they're studying 0.5% of the decisions they actually made, and studying them through a lens that omits 25 of the ~40 inputs a real read requires. The other 99.5% compound, unexamined, into whatever your win rate happens to be.

That's not an argument for reviewing more hands. It's an argument that the 20 you do review are being reviewed badly, because the file you're reading is a summary of a hand you played with far more information available at the time. You're grading your past self on a truncated exam.

Where this leaves the study session

The interesting question isn't whether the export format should expand — it should, and it won't, at least not from the rooms. The interesting question is whether the annotation habit can be made cheap enough to survive. A one-line note per interesting hand, typed at the table, takes maybe eight seconds. Over a 500-hand session that's under an hour a week, and it captures the timing, the image and the rejected action that the file will never hold. Whether that's a real solution or just a more disciplined way of papering over a format designed for the wrong job is, honestly, still open — and worth testing against your own next 1,000 hands before you decide.