Note-Taking That Wins Money: Turning Reads Into Repeatable Exploits

Most poker notes are useless because they describe a player instead of telling you what to do. Here's how to write notes that fire a trigger and change a future decision.

A note that doesn't change a future decision is just typing. You felt productive, the database filled up, and the next time you faced that player you scanned a wall of labels and did exactly what you would have done anyway. The note cost you tank-time and bought you nothing.

The fix is a mindset shift. A good note is not a description of a player — it is a pre-written exploit hypothesis: a specific trigger that, when it fires at the table, tells you to deviate in a specific direction. "Plays loose" is a description. "Limp-calls preflop, then leads the turn with any pair; overfolds river to 75%+ bets" is an exploit you can execute the moment the spot appears.

This guide is about writing the second kind and ignoring the first.

The core test: would this change a decision?

Before you log anything, ask one question: the next time I'm in a pot with this player, does this note tell me to do something different?

If the answer is no, don't write it. "Won a big pot with AA" — useless, everyone has aces sometimes. "Showed down 95s after cold-calling a 3-bet from the BB" — now we're talking, because that's a range read that bleeds into a dozen future decisions.

The whole discipline collapses to that one filter. Notes are expensive: they take attention to write and attention to read, and a database full of noise hides the three reads that actually matter. So the bar is high. A note earns its place only if it changes how you'll play a future hand.

Anatomy of an exploit note: trigger plus action

Structure every meaningful note as two halves.

Compare:

| Bad note (description) | Good note (trigger → action) | |---|---| | "Aggressive." | "Triple-barrels rivers as a bluff with missed draws; showed 76s air on KQ2-8-3. → Hero-call rivers wider, especially when I block value." | | "Tight player." | "Folds BB to 2.2x opens ~70% of hands. → Open wider/smaller from late position vs him; don't 3-bet bluff, he's not opening light to begin with." | | "Calls too much." | "Calls down with any pair, never folds top pair. → Value bet thin three streets, zero bluffs into him." | | "Tricky." | "Min-check-raises flop as a bluff on dry boards; gives up turn. → Float flop, take it away on turn." |

The right-hand column is sound exploit poker. If a player overfolds rivers, you bluff more. If a player never folds a pair, you stop bluffing entirely and value bet thinner than you would versus a thinking opponent. The note's job is to point you at the correct deviation, not to make you feel like an analyst.

Which datapoints actually carry signal

Not all observations are equal. Here's the hierarchy, from gold to garbage.

Showdowns — the gold

A showdown is the only time you see the actual hand attached to the actual line. Everything else is inference; a showdown is data. When a player turns over their cards, you get a free, verified sample of how their range maps to their betting.

What to capture: the full line plus the holding. Not "had 95s" but "limp-called BTN, check-called flop, led turn, check-called river with 95s on 9-high board." That tells you their limping range contains junky suited connectors, that they lead turn with marginal made hands, and that they don't fold middle pair. Three exploitable reads from one showdown.

Showdowns where the player's line was surprising are worth ten ordinary ones. A standard value line with a standard hand teaches you nothing. A weird line — a check-raise that turned out to be air, a passive call-down that held the nuts — is a window into how they think.

Bet-sizing tells

Many players, especially live and in soft online pools, have sizing patterns that are specific to them. The classic: small bet = weak/thin, large bet = polarized (nuts or air). But the direction isn't universal — some recreational players do the opposite and bet big when they're scared and want you gone, small when they've got it and want a call.

The note has to be player-specific and direction-explicit:

A sizing read is only an exploit when you've tied a sizing to a strength for that specific player. "Bets big sometimes" is noise.

Preflop tendencies

Preflop reads are high-frequency — they fire every orbit — so they compound fast. The ones worth logging:

Stage-specific behavior (the MTT edge)

This is where tournament players print. The same opponent is a different player at different stages, and ICM warps everyone.

A stage tag on a note matters because the read is conditional. "Nitty" without a stage is half a note.

The note-spam trap

The most common mistake among players who do take notes is taking too many. Every fold, every standard line, every "hmm" gets logged, and within a month the database is a swamp. When you finally face a tough decision and pull up their profile, you're skimming fifteen entries to find the one that matters — and you miss it.

Notes have a signal-to-noise ratio, and volume kills it. Three sharp, exploit-shaped notes beat thirty vague ones. The discipline is subtraction: log the surprising, the exploitable, the stage-specific. Skip the standard. If a player did the textbook thing in a textbook spot, that's not a read — that's just poker.

A useful rule: if you can't write the action half, don't write the note. "Played a hand kind of oddly" has no action attached. Either you can articulate what you'll do differently, or it goes in the bin.

A tagging and shorthand system

You don't have time to write paragraphs mid-tournament. Build a shorthand so a read takes five seconds to log and one second to read back. A simple, consistent system:

Examples:

The ? → ! promotion is the heart of it. One showdown is a hypothesis, not a law. The second confirming sample promotes it. This protects you from the classic blunder of stacking off light because of a single hand that might have been a one-off. In shadepoker's Player Notes tool this maps cleanly: keep the structured trigger-and-action read on the player's profile, flag confidence, and update the tag from ? to ! when the pattern repeats — so the read you pull up mid-hand is already shaped as an instruction, not a paragraph you have to decode under a 20-second shot clock.

Online vs live note-taking

The medium changes the constraints.

Online, you have a HUD-adjacent advantage: you can type freely between hands, you're multi-tabling so reads are short-lived per opponent, and players reload into your pool constantly. The trap online is volume — easy typing leads to note spam. Stay ruthless. Online pools also rotate fast, so the individual read decays; for most opponents you'll have a tiny sample and should lean on population tendencies (more on that below).

Live, you can't type at the table without it being weird, and you certainly can't pull up a database mid-hand. So live note-taking is memory-first, logged later — after the session or on a phone in a break. The upside: live samples are deeper. You play the same person for hours, sometimes days across a festival, and physical/behavioral tells layer on top of betting patterns. A live note like "stares at the board when bluffing, looks away when value-betting" is a real, repeatable read you'll never get online. Log the betting patterns the same structured way; just accept the workflow is delayed.

One discipline that transfers both ways: log the read close to when it happened. A showdown you meant to note three orbits ago is half-remembered and half-invented by the time you type it.

Population reads when you lack a sample

Here's the part most players skip: most of the time you won't have a meaningful individual read. New table, new opponent, online pool churning, three hands of history. So what's your default?

Your default is the pool tendency — the way the population at your stake and format plays, as a baseline you apply until an individual read overrides it. Population reads are the floor; individual notes are the override.

Mid-stakes online MTT pools, for example, tend to:

These aren't reads on a person; they're reads on a type of game, and they're your operating assumption against the unknown. The moment a specific player gives you a showdown that contradicts the pool — say they show up bluffing rivers in a pool that under-bluffs — the individual read wins, and you adjust just for them.

The hierarchy:

  1. No info → play the pool tendency.
  2. One surprising sample → form a hypothesis (?), lean toward it cautiously.
  3. Confirmed pattern → exploit firmly (!), even against what the pool usually does.

This also keeps you safe from leveling yourself. Don't invent a deep read from nothing and deviate hard off a hand that was probably standard. Absent real signal, the pool tendency is almost always more reliable than a story you spun from one orbit.

Turning a note into an in-game adjustment

The payoff moment: the trigger fires, and you execute. A worked example.

Your note reads: SD! check-call flop & turn, then leads river only when made a hand; gives up otherwise.

You're on the river. Two streets of check-calls, then he leads into you. The trigger has fired — he leads the river only when he's got it. The action is automatic: you fold everything but your own strong holdings; you do not bluff-raise (he's not folding), and you do not pay off your bluff-catchers (his lead is value-weighted). One showdown, logged as a trigger-and-action, just saved you a stack you'd otherwise have donated by treating his lead as the merged range a solver would assume.

Contrast the version where the note said only "tricky." You'd be guessing. The structured note does the thinking in advance, so at the table — where you have seconds and a tired brain — you're executing a pre-made plan, not solving from scratch.

The takeaway

A note is only worth taking if it changes a future decision. Everything else follows from that:

Stop describing players. Start writing exploits you can execute. The database isn't a diary; it's a list of pre-made decisions waiting for their trigger to fire.