Here, f means the probability that every opponent folds to the bet. The fold branch contributes f × P to bet EV, where P is the pot before your bet. That contribution is not the entire EV of the bet and is not necessarily the gain over checking. The called branch still matters, including how much you risk and your equity against the range that continues.
Fold Probability and Fold Equity
Suppose the pot is $100 and you expect folds 40% of the time. The fold contribution is 0.40 × $100 = $40. This does not mean that betting wins $40 on average overall. If the remaining 60% of outcomes lose a $100 bet, total EV is $40 − $60 = −$20. A positive fold contribution can coexist with a losing bet.
Keep the units visible. “40%” is a probability; “$40 of EV from the fold branch” is a monetary contribution in the stated $100 pot. “This bet gains $4 over checking” compares complete action values. All three can describe one model, but they are not interchangeable definitions. An unlabeled “fold equity: 40” is not enough to audit a hand.
The examples below use incremental EV from the point before Hero bets. Earlier contributions are sunk, the current pot is available to win, and an uncalled bet is returned. Unless stated otherwise, there is one opponent, no rake, no ties, no raises, and no future betting after a call. They are original toy models, not observed player frequencies or solver recommendations.
A Pure Bluff With No Showdown Equity
For a pure bluff that always loses when called, the two outcomes are simple: win the existing pot P when the opponent folds, or lose the bet B when called. EV(bluff) = fP − (1 − f)B. Setting EV to zero gives the break-even fold rate B/(P + B). Upswing's risk-to-reward section uses this same pot-before-bet convention.
Take P = $100 and B = $50. At 35% folds, EV = 0.35 × $100 − 0.65 × $50 = +$2.50. At 30% folds, EV = $30 − $35 = −$5. The exact break-even point is 50/(100 + 50) = 33⅓%. At that frequency the bet is neutral in this model; profit requires a rate above the threshold, not merely equal to a rounded “33%.”
The denominator is P + B, not P + 2B. The latter would be the final pot if the opponent called, but this zero-equity bluff never wins that pot. The bluff's successful reward is the old pot; its failure costs the bet. The caller's pot odds answer a different question and therefore use a different denominator.
Break-Even Fold Rates by Bet Size
The original table normalizes the pot to 100 units. It assumes a pure bluff that loses whenever called, with no further actions or costs. It reports mathematical thresholds, not predictions of how often a population folds. A 200%-pot bet needs more folds to cover its larger risk; whether it actually creates those folds is an additional range-and-response question.
| Bet Size Relative to Pot | Bet Into 100 Units | Break-Even Fold Rate, B/(P+B) | EV if Folds Are 40% |
|---|---|---|---|
| 25% Pot | 25 | 20.00% | +25.00 |
| One-Third Pot | 33⅓ | 25.00% | +20.00 |
| 50% Pot | 50 | 33.33% | +10.00 |
| 75% Pot | 75 | 42.86% | −5.00 |
| 100% Pot | 100 | 50.00% | −20.00 |
| 150% Pot | 150 | 60.00% | −50.00 |
| 200% Pot | 200 | 66.67% | −80.00 |
The last column deliberately holds fold frequency fixed. It shows that more risk is not automatically more profitable. It does not claim that a quarter-pot bet gets the same folds as a two-pot shove. To compare sizes in a real review, assign each one its own response range and justify the difference.
Add Equity When Called
A semi-bluff can win after getting called, so its called branch is not an automatic loss. Crucially, the equity input belongs against the calling range, not against every hand that reached the node. SplitSuit's fold-equity calculator guide makes that conditional-range distinction explicitly. Hands you expect to fold should not also be included as though they call. (Open the calculator)
With e as Hero's equity conditional on a call, the full single-street model is EV(bet) = fP + (1 − f)[e(P + B) − (1 − e)B]. If called and winning, Hero gains the old pot plus the opponent's call: P + B. If called and losing, Hero loses B. The same called-branch value can be written e(P + 2B) − B. Use one equivalent form, not pieces of both.
For P = $100, B = $50, f = 20%, and e = 20%, the fold contribution is $20. The conditional called EV is 0.20 × $150 − 0.80 × $50 = −$10. Because a call happens only 80% of the time, its contribution to total EV is 0.80 × −$10 = −$8. Total bet EV is therefore +$12.
The incorrect shortcut would add “20% folds + 20% equity” and call the result 40% total equity. That double-counts incompatible conditions. The chance of being called and then winning is 0.80 × 0.20 = 16%, not 20%. The mutually exclusive final outcomes are 20% folds, 16% called wins, and 64% called losses. Their weighted net results reproduce $20 + $24 − $32 = $12.
Let A denote the conditional called EV, e(P + 2B) − B. If A is negative, break-even f solves fP + (1 − f)A = 0, so f = −A/(P − A). In the $100/$50 example with 20% called equity, A = −$10 and f = 10/110 = 9.09%. If the called branch is already nonnegative, no folds are required just to reach zero EV in this simplified bet-versus-zero comparison.
| Equity When Called, P=100 and B=50 | Conditional Called EV | Fold Rate Needed to Reach Zero Bet EV | Bet EV at 20% Folds |
|---|---|---|---|
| 0% | −50 | 33.33% | −20 |
| 10% | −30 | 23.08% | −4 |
| 20% | −10 | 9.09% | +12 |
| 25% | 0 | 0% | +20 |
| 30% | +10 | None Needed for Nonnegative EV | +28 |
This is not permission to bet every hand above a called-equity threshold. Checking can be worth more, the opponent can raise, and future betting can alter the payoff. “No folds needed to beat zero” is weaker than “betting beats checking.” Keep the comparison action visible, especially when your hand has showdown value.
Why More Folds Do Not Always Mean More EV
The value of generating a fold depends on what the folded hands could have done against you. GTO Wizard's experiment shows why folding out hands with no value against your hand may add nothing relative to checking, while denying live equity can matter. A fold contribution alone does not measure the opportunity cost of losing a value call.
Here is a clean original counterexample. You hold the unbeatable hand on the river in a $100 pot. You bet $50; the opponent can only call or fold, and no rake applies. If folds occur 40%, bet EV is 0.40 × $100 + 0.60 × $150 = $130. If folds rise to 70%, EV becomes $70 + $45 = $115. You win the pot more often without showdown, yet earn $15 less because fewer worse hands pay $50.
Contrast that with the pure bluff. Holding pot, bet size, and zero called equity fixed, more folds always improve its EV: each extra fold replaces a losing called outcome with a winning uncontested one. The error is extending that statement to every hand or every size. Changing size can change the continuing range, and changing the range can change called equity.
A hand with showdown value also has a checking alternative. On the river, folding out a hand you already beat does not improve your checkdown result unless other action possibilities matter. On earlier streets, even a currently worse hand may have live outs, so denying it a free card can change value. State the street and the counterfactual line before calling a fold “extra profit.”
Multiple Opponents and Future Streets
In a multiway pot, winning uncontested requires every opponent to fold. The probability is not the average of their individual fold rates. In an original example, opponent A folds 60%, and opponent B folds 70% conditional on A having folded. The all-fold probability is 0.60 × 0.70 = 42%. The conditioning matters because B sees A's action and the players' possible hands interact through known cards.
Lexy Gavin-Mather reviews a 17 BB tournament hand in English, with English captions, illustrating two opponents folding. This is a contextual hand example, not a universal KQo shove recommendation or a solved ICM range; it differs from the cash-game EV model above.
Do not multiply two unconditional statistics unless independence is an explicit, defensible assumption. Two opponents who each fold 70% in isolation do not automatically give you a 49% all-fold rate in this pot. Their ranges, positions, action order, and the prospect of a caller behind can change the response. A multiway model should name the joint branches instead of borrowing a heads-up fold figure.
For a tightly bounded arithmetic example only, if all-fold frequency is 42%, P = 100, B = 50, and Hero always loses B whenever anyone continues with no further betting, EV is 0.42 × 100 − 0.58 × 50 = +13 units. If Hero can win after a call, or two callers create a different payoff, those outcomes need their own rows. The original two-branch simplification no longer describes the hand.
The same limitation applies across streets. A turn semi-bluff is not an all-in model if the opponent can raise now or bet the river. A draw that misses might bluff again, check and lose, or occasionally win unimproved. A draw that hits might collect more or lose to a redraw. Include those future costs and receipts before interpreting the bet's complete EV.
Range Assumptions and Sensitivity
For a post-session estimate, list the hands that arrive at the node, then separate folds, calls, and raises for the proposed size. Apply card removal before counting combinations. If a combination sometimes folds, use that frequency rather than treating it as always folding. Divide the weighted folding combinations by the weighted total that reached the node; do not divide by the full preflop range after later actions have removed hands.
In a deliberately simple model with 24 equally weighted legal combinations, six folds give a 25% fold rate. If eight combinations fold, the rate is 33⅓%. A $50 pure bluff into $100 is losing in the first model and break-even in the second. To exceed the threshold at the same equal weights, more than eight combinations must fold. Counting “eight likely folds” is not enough without the denominator.
Minimum defense frequency, or MDF, is often written P/(P + B), the complement of the zero-equity bluff's break-even fold rate in this simple setup. For a half-pot bet it is 66⅔%. GTO Wizard's “MDF & Alpha” gives the initial-bet formula and explains its limitations. It is a model benchmark, not a measured pool tendency, a requirement that every individual hand continue, or an automatic multiway strategy. The GTO poker guide provides the broader equilibrium context; do not infer a real opponent's response from the benchmark alone.
Test the close inputs. With P = 100 and B = 50, changing a pure bluff's fold estimate from 30% to 35% moves EV from −5 to +2.5 units. If your read cannot distinguish those frequencies, the correct review conclusion is “sensitive to the folding assumption.” An exact calculation can identify the uncertainty without eliminating it.
Review a Bluff After the Session
Rebuild the pot, effective stacks, action order, and exact board before discussing the bluff. Specify the proposed size, list the continuing range, and calculate equity only against the hands that call. Then build final branches with their net outcomes. Check that probabilities sum to 100% and that your uncalled bet has not been counted as an extra prize.
Compare the complete EV with checking or another legal line. A profitable bluff relative to giving up may still earn less than a smaller size; a hand with showdown value may prefer checking. Use poker ranges to make the input explicit and the hand-review process to separate the decision from its result.
The useful takeaway is not “aggression gives you two ways to win.” It is a specific conditional statement: which hands fold, which hands continue, how your hand performs when called, and whether the complete line outperforms the alternative. That is the difference between a fold-rate guess and an auditable bet-EV calculation.
Sources and Calculation Notes
- Upswing Poker — “What Is Fold Equity and Why Does It Matter?”, “What Is Fold Equity?” and the risk-to-reward formula. Supports the probability convention and pure-bluff threshold. No publisher date was visible in the fetched article body. Checked October 28, 2026.
- GTO Wizard — “The Value of Fold Equity — Experiment” (July 4, 2022), “Overview,” “Flop Analysis,” and “Conclusion.” Supports the terminology caveat and value relative to checking. Its experiments are not presented here as population folding data. Checked October 28, 2026.
- SplitSuit — “What Is a Poker Fold Equity Calculator?”, especially the input list and explanation of equity against the calling range. Supports that conditional input only; none of its worked numerical examples are reused. No publisher date was visible in the article body. Checked October 28, 2026.
- GTO Wizard — “MDF & Alpha” (July 1, 2022), “The Formula,” “Limitations,” and “When Checking Back Has Equity.” Supports the initial-bet defense benchmark and its limits. Checked October 28, 2026.
Every threshold table, EV result, weighted-combination example, multiway branch, and value-bet counterexample here is an original calculation with explicit assumptions. No fold rate is claimed to describe a real player or pool. Strategy in a game with raises, future streets, rake, or side pots requires an expanded model.
About the Editorial Team
The Poker-Eye Editorial Team researches, checks, and maintains product and educational content. Sources and product boundaries are reviewed before publication.