A bluff can be well reasoned and still get called. A poorly supported bluff can get through. SplitSuit's bluffing guide separates the price of a bluff from the opponent's assumed folds. This article adds an exact-card comparison: the same line with two ace-jack holdings, a declared opponent range, and a second response model in which even the better blocker should check.
Start With What the Bluff Is Trying to Change
If your bet makes only worse hands fold, it has not demonstrated a bluffing benefit. On the river, those worse hands were already losing at showdown. A successful bluff changes the result by making a better hand relinquish the pot. Before the river, a bet can also deny equity to hands that currently trail but could improve, so the sources of value can overlap.

SplitSuit derives risk divided by risk plus reward and compares half-pot, two-thirds-pot and pot-size bets in English, with English captions. These are immediate pure-bluff thresholds with no showdown return, not forecasts of an opponent's folds or complete multi-street EV.
A pure bluff has little or no winning equity when called under the range being analyzed. A semi-bluff also has a plausible route to improving into a winning hand while cards remain. Those labels depend on the holding, board, and opponent range. Ace-high can be a bluff against pairs but have showdown value against missed lower draws; “no pair” does not automatically mean “must bluff.”
Do not call every draw a good semi-bluff. A low flush draw can be dominated, an apparent out can complete an opponent's stronger hand, and a raise may deny your chance to see the next card. State which improvements matter, what you expect to fold, and how you respond if called or raised. The chance to get there is one input, not permission to ignore the price.
Build the Opponent's Range Before Selecting a Candidate
Begin with the positions, effective stacks, preflop action, exact board, prior sizes, and remaining players. Then distinguish hands that can arrive from hands that would actually continue against your bet. A flop call can include draws, pairs, slow-played strong hands, or other holdings depending on the strategy. It is not a guarantee that the caller has a weak pair waiting to fold.
Keep legal card combinations separate from action weights. Known cards mechanically remove impossible holdings. The chance that an unblocked holding takes a particular line is an estimate. Poker ranges gives the counting framework. Write those estimates down, especially when you are tempted to explain a line with “they cannot have anything here.”
Next, divide the proposed response into folds, calls, and raises. Ask whether the folding hands are better than yours, whether a size change alters those groups, and whether checking has value. Do not use a total fold-to-c-bet statistic as a direct river-fold estimate. Different streets, lines, sizes, opponents, and samples answer different questions.
One Fixed Line for the Blocker Comparison
The following is an original No-Limit Hold'em teaching hand, not observed play or recommended sizing. Six-max; blinds 0.5bb/1bb; no ante or rake; both active players start with 250bb. BTN raises to 3bb, SB folds, and BB calls by adding 2bb. The flop is K♠Q♠8♦. BB checks, BTN bets 6.5bb, and BB calls. The turn is 2♣. BB checks, BTN bets 40.25bb, and BB calls. The river is 3♠, and BB checks again.

| Completed stage | BTN's added contribution at that stage | BB's added contribution at that stage | Pot | Each active player's remainder |
|---|---|---|---|---|
| Preflop, 3bb each plus dead SB 0.5bb | 3bb | 3bb, including posted blind | 6.5bb | 247bb |
| Flop bet and call | 6.5bb | 6.5bb | 19.5bb | 240.5bb |
| Turn bet and call | 40.25bb | 40.25bb | 100bb | 200.25bb |
BTN now considers a 75bb river bet. Compare A♠J♦ with A♥J♦, keeping the line and proposed size fixed. The large turn overbet is simply part of the declared history; no claim is made that either holding should have taken it. We evaluate the river snapshot, not retroactively certify the earlier bets. The range below is also stipulated at the river, not presented as a realistic complete defense of that turn size.
Both holdings have ace-high on this river. A♠J♦ is not a flush: the board has three spades and the hand supplies one, for four spades in total. An opponent with two spades can have a flush. That distinction is crucial: the ace of spades can block nut-flush holdings while its owner still has no made flush.
Define the Range Before Calculating the Outcome
For this deliberately small model, give each exact holding below weight one if it remains legal after removing BTN's cards. The range contains fourteen flush holdings and twelve one-pair holdings before that removal. It is not every possible flush or pair, and equal weight is a teaching assumption. In the first response model, every flush calls 75bb, every listed pair folds, and nobody raises. Both ace-high candidates lose whenever called and also lose if they check back against this defined range.
| Group before BTN-card removal | Exact holdings | Response in model 1 |
|---|---|---|
| Seven A♠ flushes | A♠J♠, A♠T♠, A♠9♠, A♠7♠, A♠6♠, A♠5♠, A♠4♠ | Call |
| Seven other flushes | J♠T♠, J♠9♠, T♠9♠, 9♠7♠, 7♠6♠, 6♠5♠, 5♠4♠ | Call |
| Three KJ pairs | K♥J♣, K♦J♣, K♣J♣ | Fold |
| Three KT pairs | K♥T♣, K♦T♣, K♣T♣ | Fold |
| Three QJ pairs | Q♥J♣, Q♦J♣, Q♣J♣ | Fold |
| Three QT pairs | Q♥T♣, Q♦T♣, Q♣T♣ | Fold |
With A♥J♦, none of these 26 holdings is blocked by BTN's cards. With A♠J♦, the seven A♠ flushes are impossible, leaving nineteen holdings: seven calls and twelve folds. The twelve pair holdings are unchanged because none contains A♠, A♥, or J♦. Normalize each surviving range separately; do not divide both cases by 26 after removing seven impossible hands from one case.
Using the decision-point convention, a fold wins the existing 100bb pot and a call loses the new 75bb. Checking returns zero in this particular model because every opponent hand beats ace-high and checking closes the action. Thus bluff EV = fold share × 100 − call share × 75. The zero check value is stipulated by the range, not a general property of checking.
| BTN candidate | Surviving calls | Surviving folds | Modeled fold share | 75bb bluff EV | Check EV |
|---|---|---|---|---|---|
| A♠J♦ | 7 | 12 | 12/19 = 63.16% | (1,200 − 525)/19 = 35.53bb | 0bb |
| A♥J♦ | 14 | 12 | 12/26 = 46.15% | (1,200 − 1,050)/26 = 5.77bb | 0bb |
The ace of spades improves the bluff comparison in this response model because it removes seven calls while leaving the folds intact. Both bluffs are positive relative to checking under the stipulated responses. This does not imply that a real opponent folds those pairs or reaches this river with these weights. The range assumption does most of the strategic work; the arithmetic only tells us what follows from it.
A Good Blocker Does Not Force a Bluff
Now change the response assumption before calculating a new result. Keep the same surviving ranges, pot, 75bb size, and no-raise rule. Every flush still calls, but each listed pair now calls half the time and folds half the time. This is model 2. We are not claiming either model describes a population; the contrast tests how much the earlier conclusion depended on the pairs folding.
| BTN candidate | Weighted calls in model 2 | Weighted folds | Bluff EV | Better of bet and check in this model |
|---|---|---|---|---|
| A♠J♦ | 7 flush + 6 pair = 13 | 6 | (600 − 975)/19 = −19.74bb | Check: 0bb |
| A♥J♦ | 14 flush + 6 pair = 20 | 6 | (600 − 1,500)/26 = −34.62bb | Check: 0bb |
A♠J♦ remains the better of the two bluff candidates, but both bluffs lose to checking. Removing some strong calls is not enough if the rest of the range continues too often. That is the difference between selecting a candidate conditional on bluffing and establishing that a bluff belongs in the strategy at all.
For a zero-equity, one-bet river bluff with no raises or later action, betting 75bb to win 100bb needs more than 75 ÷ (100 + 75) ≈ 42.86% folds to outperform a zero-value check. This is a threshold, not an estimate of the opponent's behavior. The fold-equity guide owns the full break-even calculation; here the key point is that your assumptions must supply the folds rather than the formula inventing them.
Unblocking Folds Is a Separate Question
In the first model, change A♠J♦ to A♠J♣. The ace of spades still removes the seven A♠ flush calls. But J♣ now also removes the three KJ folds and three QJ folds. Only seven flush calls and six pair folds remain. The bluff EV becomes (6 × 100 − 7 × 75) ÷ 13 = 75 ÷ 13 ≈ 5.77bb, far below the 35.53bb result with A♠J♦.
This is why players talk about unblocking the folding range: a candidate can be preferable because its cards do not remove the hands you want to fold. It is a range relationship, not a special property printed on a card. A card that removes a call in one model might remove a fold in another. Compare exact holdings, not a generic “ace blocker” label.
Also remember that real action weights may differ between the holdings before the river. A blocker can affect earlier calls, raises, and continuation choices. Our controlled comparison holds the hypothetical range model fixed except for legal-card removal so you can inspect one mechanism. It does not prove the two holdings reach this node at equal frequency in a solved strategy.
Credible Value Means Hands That Can Actually Take the Line
Ask what value hands you could arrive with through the same actions and then bet for this size. On the example's three-spade river, two-spade holdings can make flushes. But you still need to ask which of those hands raise preflop, bet the flop, make the large turn bet, and reach this river. Saying “the board has a flush” does not establish a credible value range for every player in every line.
Do not choose all possible bluff candidates independently. If you bet every missed draw because each has a nice-sounding blocker, the combined bluffing range can become much larger than intended relative to the value hands. Count the whole proposed betting range with its weights. One appealing candidate does not tell you how often the range should bluff.
Red Chip's bluffing discussion explains why polarized models depend on assumptions about value hands, bluffs, and bluff-catchers. Treat a toy model as a controlled calculation, not as a universal bluff-to-value chart. The GTO poker guide provides the broader distinction between a model solution and an opponent-specific adjustment.
Plan Future Streets Without Committing to Every Barrel
On the flop or turn, name the cards that improve your hand, the cards that strengthen your credible value range, and the cards that improve the opponent's continuing range. Then consider the response to each major branch: fold, call, or raise. A plan can include checking or giving up. Continuing merely because you already invested money is not a new reason to bluff.
For example, A♠J♦ on K♠Q♠8♦ has a gutshot to Broadway through a ten and a backdoor spade-flush possibility. The turn 2♣ does not turn that backdoor possibility into a direct flush draw. On the river 3♠, it still does not make a flush. You must update the holding at each street rather than carrying the phrase “nut-flush blocker” as if it supplied the same equity throughout the hand.
A multi-street bluff cannot be evaluated by adding the isolated success stories from each street. Later decisions happen only after earlier responses, and the range changes along the way. A flop bet that loses immediately can sometimes form part of a better complete line; it can also be an unnecessary loss followed by an even more expensive turn mistake. Expose the conditional branches and their costs instead of assuming the second barrel fixes the first.
Multiway Pots and Player-Specific Limits
If you need two opponents to fold, one player's likely fold is not enough. Their decisions also need not be independent: a bet, call, or fold from one player can change another's response. Rebuild the range and action order before applying a heads-up example. Stack sizes and side-pot rights can create additional constraints: PokerStars' pot rules specify that only contributors can win a side pot, so not every chip in the middle is available to every player.
Player-specific adjustments need relevant evidence. A tendency observed against small flop bets may not predict calls against a large river bet. A single hero call does not establish that someone never folds; one folded pair does not establish that they overfold. Preserve the sample and context behind a read, and use a cautious range of responses when that evidence is thin.
Avoid treating a pause, a quick click, a chat message, or a physical gesture as a reliable standalone detector of hand strength. Timing can have many causes, and an explanation that fits one revealed hand may fail on the next. Observable actions and well-documented patterns can inform a hypothesis, but they do not remove uncertainty about an opponent's cards.
Review a Bluff Without Results Bias
Save the action record before writing your verdict. Include the exact cards, positions, pot, stacks, sizes, remaining players, proposed range, and the hands expected to fold or continue. Then separate known inputs, assumptions, and the result. The hand-analysis guide helps keep those fields distinct.
| Before betting | After the session |
|---|---|
| Which better hands can fold this size? | Were those hands plausibly in the range before seeing the result? |
| Which exact calls does my holding block? | Did I also block hands I needed to fold? |
| What value hands take this line? | Did I count the full betting range or just this one candidate? |
| What happens if called or raised? | Did my later action follow a conditional plan or defend sunk costs? |
| What is checking worth? | Did I compare complete alternatives from the same point? |
| Which response weight could reverse the answer? | What new evidence actually changes that weight? |
Suppose A♠J♦ gets called by one of the seven surviving flushes in model 1. That outcome is already allowed by the positive-EV model; it does not refute the calculation. If A♥J♦ gets a fold in model 2, that outcome is also allowed by a negative-EV model; it does not vindicate the decision. What matters is whether the range and response assumptions were supported before the outcome became known.
The most useful review can end with “the decision is sensitive to whether these pairs fold.” That is more informative than “great bluff” or “terrible bluff” based on one pot. It identifies the evidence you need next and prevents a blocker, a scary river, or a successful result from becoming an automatic instruction.
Sources and Model Boundaries
- SplitSuit — “Poker Bluffing 101 (Finding +EV Bluffs)”, reviewed October 28, 2026. Poker Bluffs and Breakeven %, Will They Fold When I Bluff?, and Double Barrel Bluffs support separating risk-reward thresholds, response assumptions, and later streets. Its illustrative percentages and player tendencies are not transferred to this example.
- Red Chip Poker — “Mastering the Art of Poker Bluffing: Strategies, Tips, and GTO Analysis”, reviewed October 28, 2026. Assumptions and Simplifications in Game Theory and the polarized-range discussion support the importance of model boundaries. No universal “perfect bluff” frequency or numerical example from that article is adopted.
- GTO Wizard — “Reasons for Value Betting in Poker”, December 6, 2022; reviewed October 28, 2026. Sources of Value explains semi-bluffs and overlapping sources of value before the river. No solution figures are imported.
- PokerStars — “The Rules of Poker”, reviewed October 28, 2026. Table Stakes and All-in and The Pot support the side-pot eligibility caveat, not a bluffing frequency.
The fixed hand, exact opponent holdings, equal weights, call/fold assumptions, chip ledger, blocker comparisons, and EVs are original educational models. They are not observed player data, solver output, or a product performance test. Changing the range or its response can reverse a conclusion, including when the same blocker remains in your hand.

