Hand history review without results bias

Stop the hand at the decision you want to study. Hide the later action, river and showdown. Write down what the record establishes, what you are estimating and why you favor a particular choice. Reveal the ending afterward, with that earlier note still beside you. A result is useful evidence, but a lost pot cannot identify an earlier mistake by itself.

This lesson gives you a repeatable review sequence and a small fictional exercise. Two cases have identical turn information and different endings. A third has missing amounts. The aim is to leave a clear account of your reasoning, including the parts you cannot yet defend. There is no solver verdict, strategy score or claim that completing a worksheet improves your win rate.

Indira: “Before we open the ending, what are you trying to decide?” Ace: “Whether the turn call made sense at the price we were offered.” Indira: “Keep that question. A later river card does not answer it on its own.”

Choose a decision · Separate facts and estimates · Check the paired turn · Capture a review · Use the later record

Choose one decision before opening the ending

“Was this hand bad?” gives the reviewer too much room to wander. Choose a point where a player had an actual choice. “Did the turn call meet the required price under my range estimate?” has identifiable inputs. “Should I bet the river for value against this calling range?” is a different question and needs its own information.

Mark the exact street and the player whose decision you are reviewing. Record the game variant, positions, effective stacks and action through that moment. Include the amounts already committed and say whether a pot figure includes the latest bet. If the question involves a later street, preserve the earlier action that led there rather than showing a board and asking for a guess.

If you remember the ending, add “result already known” to the note. Concealing a panel cannot erase that memory. You can still write a disciplined review, or ask someone to prepare a different hand extract. The exercise is a visible information boundary for study, not a secure blind experiment. The page source and no-JavaScript explanation contain the endings.

Keep facts, estimates and gaps under separate labels

A reliable record can establish that an opponent bet 10 into 30. It cannot establish the range you assigned to that bet unless you recorded that estimate separately. “The opponent bet 10” and “I assumed a range against which I win 25% of the time” are different kinds of statement. Both matter, but only the first is a recorded action.

An assumption should have a reason and a boundary. Position, earlier sizing or a clearly described observation may be part of its basis. Avoid turning “I had no read” into a confident story about the opponent. If an estimate comes from a study model, identify the model and its inputs. The exercise below simply supplies an estimate so you can examine the review process.

Leave missing data visibly missing. A copied hand may omit the current pot, a raise-to amount or the shorter stack. Do not insert the number that would make the hand look familiar. Ask for the original record or narrow the question to what the extract supports. A tidy calculation built around an invented bet size is still built around an invented bet size.

Read the paired turn before doing the arithmetic

Cases A and B use fictional heads-up no-limit Hold’em. Hero holds ace of hearts and king of hearts. The board is queen of hearts, seven of hearts, two of clubs and three of diamonds. The pot contains 30 units before the opponent bets their final 10. Hero has 20 units behind and faces a call of 10. These are chip units, with no currency or stake claim.

The opponent is all-in, so no further betting follows a call. Hero can call 10 or fold; the unused 10 in Hero’s stack cannot create an additional contested wager against an all-in opponent. Earlier contributions are already in the pot of 30. We are studying this turn choice without reconstructing an unsupported preflop or flop story.

For the exercise model, ignore rake and ties and use a supplied 25% chance of winning against an unspecified opposing range. That number is not calculated from the pictured cards and is not asserted to be equity against the particular hand revealed later. Do not replace the range estimate with exact-hand knowledge before you have reached the reveal.

The pot currently contains 40, including the opponent’s bet. Calling 10 makes the final contested pot 50. The break-even win share is 10 ÷ 50 = 20%. A useful first check is whether your denominator includes your call. Using the 40 already visible would give 25%, which describes a different ratio and overstates the threshold for this simplified call.

Relative to folding now, a win brings a net 40 and a loss costs the new 10. With the supplied model, EV(call) = 0.25 × 40 − 0.75 × 10 = +2.50 units. Money committed earlier is already in the pot; it is not an extra new loss to subtract at this decision. Folding is the zero reference for this comparison, not a claim that the whole hand cost nothing.

The conclusion depends on the estimate. At 20% the same calculation gives zero; at 15% it gives −2.50. Those checks isolate the disagreement. If someone rejects the call, ask whether they dispute the pot count, the model boundaries or the estimated win share. The arithmetic does not settle whether the opposing range was assigned sensibly.

Write the objection while the result is still hidden

Try an initial note such as “The supplied 25% exceeds the 20% threshold. Calling is positive under that input, but I need a basis for the range estimate.” It contains a calculation, a conditional conclusion and a useful next question. “Easy call” hides all of those dependencies. So does “I had a draw,” which never connects the hand to the price.

Then state the strongest objection you can make without opening the ending. Here, the estimate may not represent the range that reaches this turn and chooses this bet. In another hand, the objection could concern rake, future action or a missing stack size. Writing it before the reveal gives you a fairer record of what you thought was uncertain.

Ace: “I have a positive number. Am I finished?” Indira: “You have checked the calculation. Now write why you trust the estimate, or admit that this example supplied it. That is the part the river cannot repair for you.”

Capture your review before revealing the later record

Choose Case A, record a decision, then fill in the assumption and evidence fields. Capture my review keeps that note beside the exercise and enables Reveal later record. Nothing is graded. You can disagree with the supplied model, provided your note makes the disagreement clear. For the incomplete case, identifying the missing amounts is a valid stopping point.

Editing the decision or either text field clears the captured note and conceals the later record. Changing cases also clears the inputs. This prevents words typed after a reveal from appearing as an untouched earlier judgment. It does not claim that you have never seen the outcome. Repeating a case is a comparison exercise, not a fresh blinded trial.

Reset exercise returns to Case A with empty notes and the ending concealed. The notes exist only in the current page view; this exercise does not save them, send them to an account or carry them across a reload. Use fictional text. If you want to keep a study conclusion, copy only the necessary reasoning to your own record and leave out unnecessary personal details.

After the reveal, identify which information changed

The paired endings are explained here as well as in the exercise, so complete your initial note before reading on if you want to try the sequence. In Case A, the opponent shows queen of clubs and queen of diamonds. The river is five of hearts. Hero has an ace-high heart flush; the opponent has three queens. Hero wins this constructed ending.

Case B shows the same opposing queens, but the river is queen of spades. The opponent makes four queens and wins. Every card is unique within each fictional deal. These are deliberately selected endings, not a random simulation or a sample of outcomes from the supplied range model. Their purpose is to hold the earlier information still while the result changes.

In both cases the initial pot, call price and supplied range estimate were identical. The +2.50 calculation from those inputs therefore remains identical. Winning Case A does not establish that 25% was a sound estimate. Losing Case B does not make the subtraction wrong. Both cases leave the same question about the basis for that estimate.

The shown hand does provide new information. You can ask whether queens belonged in the initial range and how the action fits that part of the range. An equity calculation against the revealed queens would answer an exact-hand question. It would not prove that you should have assigned that exact hand with certainty before seeing it. Keep the two analyses under their own labels.

A corrected input deserves a revised calculation. If the source record later establishes that the call was 15 rather than 10, preserve the first note, correct the price and calculate again. That revision has a factual reason. If only the river changes, explain why the initial model remains the same. This gives a reviewer something more precise to inspect than whether you feel better about the hand.

An incomplete hand can end with a request for evidence

Case C begins on a flop of queen of spades, eight of clubs and two of hearts. Hero holds nine of spades and nine of diamonds. The opponent bets, but the copied record contains no pot, bet size or remaining stacks. There is no supplied range estimate. You know the cards shown; you do not know the call price.

A useful note asks for the missing amounts and action context. It might say “I cannot compare the call cost with the contested pot from this extract. Obtain the pot before the bet, the bet size and effective stack.” The exercise reveals that no ending is recorded. There is no hidden correct action or secret showdown that could rescue the missing inputs.

Use a review partner without giving away the ending

For a study group, one person can prepare the extract and another can review it. Remove the result, later cards and comments that announce whether Hero won. Preserve positions, sizes and stacks needed for the question. Read the introduction too: “Here is my worst river beat” gives the reviewer information even when the showdown panel is closed.

Remove names, account identifiers, private chat and unrelated personal information before sharing a real hand through your own study process. A seat label such as Hero or Button can preserve the poker role. Check whether the record and your study group permit sharing. This page needs none of that material; its exercise uses fictional cases and offers no hand-history upload.

Keep a record you can revisit without rewriting it

A reusable note needs the decision point, source record, confirmed facts, assumptions, calculation and unresolved question. Add the later information separately. Retain enough of a date or private identifier to find your original hand, without copying unrelated details into every worksheet. When an input is corrected, keep the correction visible beside the original note.

For Cases A and B, the next question concerns evidence for the supplied range estimate. For Case C, it concerns missing amounts. Either task can be finished away from the table. “Play until I win this spot” does not check the estimate or recover the hand history. A review can end with an unanswered question and a clear plan to obtain the evidence.

Ace: “The ending changed my mood. Should it change the note?” Indira: “Add the ending on a new line. If you revise the analysis, name the new fact that justifies it. Keep the earlier note where you can still read it.”

Check the selection as well as the individual hand

If every selected hand is a large loss, the review set says more about your selection rule than about all your decisions. Write down how you chose the hands. You might examine every occurrence of one action in a bounded session, or a small offline sample chosen for a particular study question. The point is to know what the set represents before drawing a wider conclusion.

Outcome bias is a documented reason to be careful about this sequence. In five studies reported by Jonathan Baron and John Hershey in 1988, favorable outcomes affected evaluations of decisions under uncertainty. The study was not a test of this poker exercise. It supports caution about outcome knowledge, while leaving this worksheet’s effect on poker performance unestablished.

Finish each review by naming the next verifiable question. You may need a corrected hand history, a better specified range or a calculation under different assumptions. Keep that question beside the result instead of asking the result to stand in for the work. Further calculation lessons are available in Info & Tools.

Back to top