Almost everyone who reviews hands believes they are learning from them. The honest version of the experience usually looks different: you open the replayer, watch the hand unfold, decide you “should have folded the turn,” close the window, and change nothing. A week later the same spot arrives and you call again.
The gap is not effort, it is method. A review that ends with a verdict has produced an opinion. A review that ends with a conditional instruction has produced a change. The five steps below are built around that distinction, and they are followed by one fully constructed worked hand — street by street, node by node — so you can see where the value actually comes from and what to write down.
Why most hand reviews change nothing
Three failures account for almost all of the wasted effort.
- You review the hand instead of the decision. A hand is not one event. It is four to six separate decisions separated by new information, and grading them together tells you nothing you can act on.
- You review with information you did not have. Once you know the river card, your memory of the turn quietly contains it. This contamination is invisible from the inside and it is fatal to the exercise.
- You review the ones that hurt. Large pots and ugly runouts are vivid, so they get reopened, while the genuinely instructive hands — small, uncertain, unremarkable — never get loaded at all.
Each of the five steps closes off one or more of those holes.
The five-step method
| Step | What you do | What it produces |
|---|---|---|
| 1. Fix the frame | Within 30 minutes of the session ending, record the hand in raw form: positions, effective stacks in big blinds, players dealt in, opponent tendencies, your own state. | A record that can be reconstructed later without leaning on memory. |
| 2. Mark the decision nodes | Draw a line at every point where new information arrived and you had to act: preflop facing action, the flop bet, the turn barrel, the river shove. | A map showing which node actually decided the hand. |
| 3. Build ranges from the other side | At each node, write what the opponent’s line contains — and separately, what your line looks like to them. Only then open an equity calculator. | Range estimates you can check, rather than impressions. |
| 4. Grade the decision, not the result | Score each node against the information that existed at that moment. One sentence of reasoning per grade. | A verdict per node, with the reasoning attached. |
| 5. Extract one trigger and file it | Convert the worst node into an if/then instruction phrased the way you would say it to yourself mid-hand. | A rule that can fire while the clock is running. |
Step 1: fix the frame before memory rewrites the hand
The raw record is deliberately boring. Blinds, my position, their position, effective stacks, how many players were dealt in, what I knew about the opponent’s frequencies, and one sentence on my own state. That last field matters more than players expect: if I was three hours into a session and frustrated, that belongs in the record, because it is part of the information the decision was made with.
Step 2: mark the nodes before you analyze anything
The node map is what separates an audit from a retelling. Most hands contain exactly one decision that mattered, and it is usually not the last one. On the river, decisions are often obvious because the information is complete. The interesting decision happened a street earlier, when the picture was thinner and the choice was closer. Write the nodes in order, mark the one that felt uncertain, and start there.
Step 3: build ranges from the other side
Write the opponent’s range in words first — “overpairs, sets, a couple of flush draws, and a small bluff portion” — then convert it to a rough percentage of hands. The discipline of doing it in that order prevents the most common form of self-deception, which is reverse-engineering a range that happens to justify what you did. Then do the same for your own line: if your line looks strong and they bet anyway, that signal is worth more than the size of the bet.
Step 4: grade the decision, not the result
Four grades are enough. Sound: the choice you would make again with the same information. Thin: defensible, but the margins are narrow — flag it. Leak: a choice that gives up equity against the range you assigned. Blunder: a choice made with no coherent reason at all. Grade before you look at the outcome, and if your software shows results inline, cover that column on the first pass.
Step 5: extract one trigger and file it
A trigger is a conditional with a test inside it. “Be more careful with draws” is an intention and it will not survive contact with a real hand. “If I hold a draw with no pair on the turn, facing a barrel from a player whose river barreling is narrow, and my clean outs times two is below the required equity, fold” is a trigger. The second one is usable under time pressure.
Worked example: K♠Q♠ from the hijack
This hand is constructed for teaching rather than taken from a session; the structure is the point, not the specifics. Six-handed ring game, 100 big blinds effective, a recurring opponent on the button I have logged across many sessions as tight-aggressive. Their button 3-bet frequency in this configuration is roughly 8%. My notes say their turn barrels are value-heavy and their river barreling narrows sharply on flush-completing cards.
Preflop. I open K♠Q♠ from the hijack to 2.5 big blinds. The button 3-bets to 8. Both blinds fold. Pot: 12. I call.
Flop. 9♠ 4♦ 2♠. I check. Villain bets 4 into 12 — about a third of the pot. I call. Pot: 20.
Turn. 7♥. I check. Villain bets 11 into 20, a little over half the pot. I call. Pot: 42.
River. J♦. I check. Villain moves all in for my remaining 74 big blinds. I fold.
| Node | Action | Equity required | Equity I held | Grade |
|---|---|---|---|---|
| 1 | Preflop: I open, button 3-bets, I call | — | — | Sound, flagged |
| 2 | Flop: check, one-third pot bet, call | ~20% | ~26–28% | Sound |
| 3 | Turn: check, ~55% pot bet, call | ~26% | ~18% | Leak |
| 4 | River: check, all in, fold | ~39% | Near zero | Correct but irrelevant |
Node 1 — preflop. KQs against a tight button 3-bet range is a legitimate flat and a legitimate 4-bet; folding is too tight. But notice what the flat buys me: a hand that mostly makes second-best pairs and non-nut draws, out of position, against a player whose range is stronger than mine. Nothing here is a mistake. It is a flag, and it becomes the second trigger below.
Node 2 — flop. Required equity: 4 divided by the 20 the pot becomes, so 20%. My holding: nine spades remain, discounted to roughly 8.5 because a couple of suited spades sit inside villain’s 3-bet range — about 17% from the flush alone. The overcards add something against AQ, AJ and the bluff portion, and almost nothing against the overpairs that dominate their range. Call it 26 to 28% overall. Clear call. Raising would mostly build a pot where my equity is static and my position stays bad.
Node 3 — turn. This is the node that decided the hand. Required equity: 11 divided by the 42 the pot becomes, about 26%. My holding is the same 8.5 spade outs, now worth 17%, and the overcards have stopped being outs against anything that fires twice. That leaves me near 18% against a requirement of 26%. The usual escape hatch — implied value when the draw completes — is closed, because this opponent checks back or folds most spade rivers. A draw that never collects an extra bet when it arrives has nothing to subsidize a short call. Grade: leak.
Node 4 — river. Required equity: 74 divided by 190, roughly 39%, holding king-high. Fold immediately. Correct, and instructive only for what it says about node 3. When the final decision in a hand is obvious, that is a strong hint the real decision happened earlier.
What the audit produced was not “call less with draws.” It was two triggers: a general one about naked draws facing narrow river barreling, and an opponent-specific one about flatting KQs out of position against a tight 3-bettor who shuts down on completing cards.
What to log: the six-field entry
| Field | Example from this hand | Why it earns its place |
|---|---|---|
| Context tag | Ring game, six-handed, recurring opponent | Lets you filter and compare entries later. |
| Node | Turn, facing a ~55% pot barrel | Names the decision, not the hand. |
| Information available | Notes said river barreling narrows on spades | Separates a real mistake from a genuine unknown. |
| Equity check | Needed ~26%, held ~18% | Turns a feeling into a number you can argue with. |
| Grade plus one sentence | Leak: called a turn barrel with a naked draw and no river extraction | Forces a verdict instead of a vibe. |
| Trigger (if/then) | If clean outs times two is below required equity and the river pays nothing, fold | Makes the review transferable to live play. |
The first five fields are the audit. The sixth is the entire point of it. Most reviews die at field five, because a grade feels like a conclusion when it is really only a description.
One practical habit: keep triggers in a holding pen before promoting them to rules. A trigger becomes a rule only after it has fired in three separate sessions and held up each time. That keeps one strange hand from rewriting a strategy that was working, and it gives a promising but unproven trigger somewhere to live instead of being discarded.
Reviewing inside a small club pool
Private-club and online-club pools change the arithmetic of hand review in two ways. The first is recurrence: you sit with the same players often enough that opponent-specific triggers stay useful for months rather than a single session, which makes writing them worth the effort. The second is noise: twenty hands against a player is a hypothesis, not a profile, and a small pool tempts you to over-read two loud results.
Two habits handle both. Tag every entry as pool-general or opponent-specific, and never let an opponent-specific trigger override a pool-general one until it has repeated. And when the club software stores hands automatically, resist reviewing everything — the constraint is not the volume of data, it is the reconstruction. Five hands audited properly will teach you more than two hundred skimmed. In a pool where you see the same faces weekly, the well-kept log also compounds, because reads you build this month are testable next month.
Where the method breaks
- Results creep back in. If you can see the runout while grading, you will grade the runout. Cover the column.
- Range inflation. Players quietly assign opponents whatever range makes their own play look reasonable. Write the range in words before opening a calculator.
- Single-hand overfitting. One observation is not a pattern. Use the holding pen.
- Selection by drama. Reviewing only large pots filters out the honest signal. Sample every decision that felt uncertain, regardless of pot size.
- Review theater. An hour of watching replays without a single written trigger is the same as not reviewing at all.
The test of a review is not whether you understood the hand afterwards. It is whether the understanding shows up six hours later, in a different hand, with the clock running and a draw in your hand that you do not like. Build the method so that it can.
This article is informational only and is not legal advice. Rules and laws relating to poker vary by location, so check what applies where you play.
