Blank-Turn Bluffs: Start With the Checked Range

You defend the big blind with K♣ 9♣. The flop comes A♠ J♥ 6♦, you check, and the button checks behind. The turn is 2♣. Is this your chance to steal a quiet pot?

In this stored solution, checking is worth 0.584bb; betting 1.8bb into 5.5bb is worth 0.134bb. The small probe gives up 0.450bb. That is 0.584 − 0.134bb at the same decision. The three clubs among your cards and the board cannot make a flush with only one river card to come.

Start with the range that checked back and the value your hand keeps by checking. Across eight chosen low turns, automatically betting small with every unpaired hand without a direct draw failed our test. Some of those hands still make useful probes, as the comparisons below show.

These are solver-library outputs for the specified model. The range calculations describe those outputs; our explanations are interpretations, not a claim to know the cause of every mix. Solver version and achieved convergence are unavailable.

The same king-high hand can have a different job

Move to Q♠ 7♥ 2♦ followed by 3♣, keeping the positions, stack depth, rake and size menu. K♣ 9♣ now uses the small probe 42% of the time. Its highest stored EV is the small bet, though its 0.033bb edge over checking is much smaller than the first hand’s mistake.

These are two complete scenarios, with different boards and flop check-back strategies. They show a change in this hand’s job; they do not isolate the effect of replacing one board rank.

How the shortcut performed on eight low turns

We tested six-max cash solutions with 100bb starting stacks, a button open to 2.5bb and a big-blind call. The pot is 5.5bb, rake is 5% capped at 4bb, and neither player bets the flop. BB has only a check at the flop root in these trees. On the turn, BB can check or bet 1.8bb, 4.1bb or 8.2bb.

Before the eight-turn EV analysis, we set this screen: the 1.8bb probe should lose no more than 0.10bb for at least 95% of eligible BB range on each board. Eligible hands have no pair, made straight or flush, flush draw, or four-to-a-straight draw, including a gutshot. Each selected turn is a low card of the flop’s missing suit.

We used two discovery flop families and six different held-out families: 1,388 exact hand rows in total. All eight boards failed the screen. The share losing more than 0.10bb ranged from 13.5% to 43.0% of the eligible range. The six held-out boards also failed individually, so the result extends beyond the opening comparison.

Where the small bluff loses more than 0.10bb
Share of eligible BB range, on a 0–100% scale. Dashed marker: the 5% failure line.

A♠ K♥ 8♦ 2♣30.1%

Discovery · 144 eligible exact hands

Q♠ 7♥ 2♦ 3♣13.5%

Discovery · 236 eligible exact hands

A♠ Q♥ 7♦ 2♣23.4%

Held out · 128 eligible exact hands

A♠ J♥ 6♦ 2♣22.6%

Held out · 124 eligible exact hands

K♠ Q♥ 8♦ 2♣20.2%

Held out · 176 eligible exact hands

K♠ 9♥ 4♦ 2♣37.6%

Held out · 180 eligible exact hands

J♠ 8♥ 3♦ 2♣43.0%

Held out · 196 eligible exact hands

T♠ 7♥ 4♦ 2♣13.5%

Held out · 204 eligible exact hands

Giving each board equal weight, the mean loss was 0.080bb. Among the eligible hands under that same weighting, the weighted 95th percentile was 0.257bb, and the largest stored gap was the 0.450bb king-high example. At a looser 0.20bb threshold, six of eight boards still failed the same 95%-of-range screen.

Changing the entire range to a simplified betting policy would also change later play. These local losses compare one action with the best available action against the stored continuations; they are not a measured win rate or the exploitability of a new strategy.

The button’s check leaves more than missed hands

On the A♠ J♥ 6♦ flop, the button retains 59.74 weighted combinations of top pair after checking, out of 114.90 before its decision, with the turn card removed from both counts. It also retains some sets and two pair. A check has filtered the range; it has not emptied it of strong hands.

After the 2♣ turn, 60.4% of BTN’s marginal range has at least a pair. The corresponding share on Q♠ 7♥ 2♦ 3♣ is 34.7%. Both players’ ranges appear below so the comparison is visible. A made-hand count is not a measure of mean equity or a proof of why a bet has its EV.

The BB grid shows the turn choice on A♠ J♥ 6♦ 2♣. The BTN grid shows its response to 1.8bb. Both use the actual ranges reaching this street, before conditioning on one specific opposing hand. Select a hand class or use arrow keys to inspect its mix; suits within one class can differ.

GTO Wizard’s probe analysis makes the related point in a different setup: a 100bb early-position raiser can check an ace-high flop and retain a strong range. Our numbers test a button opener, so they are a separate sample.

With K♣ 9♣ fixed, the small probe actually gets slightly more folds on the ace-high board: 24.2% versus 22.6%. Yet it is the costly bet there. Fold percentage alone leaves out the value retained by checking, which hands continue, and later betting.

Even the 8.2bb probe, which gets 60.7% folds on A♠ J♥ 6♦ 2♣, trails checking with K♣ 9♣ by 0.568bb. More folds do not settle the comparison. We have described the response ranges, not isolated a single causal explanation or measured opponents’ mistakes.

Leave room for the bluffs the range does use

On T♠ 7♥ 4♦ 2♣, Q♦ J♠ has no pair and no direct draw under our definition. It bets 85% in the stored strategy. Its 4.1bb bet is worth 0.492bb against 0.394bb for checking: a 0.098bb difference, just below our 0.10bb screen but relevant to a more demanding one.

Draws need a separate comparison. On A♠ J♥ 6♦ 2♣, Q♠ T♠ has a gutshot to a king and bets 88%, mostly for 8.2bb. The highest bet EV beats checking by only 0.035bb. A frequent bluff and a large penalty for checking are different findings.

We also retrieved the top-rank pairing turn for every flop. On T♠ 7♥ 4♦ T♣, Q♦ J♠ still bets 77%; on A♠ J♥ 6♦ A♣, K♣ 9♣ still prefers checking. “Paired turn” is too broad to prescribe a single action. Those eight paired-turn trees are comparisons, not eight more independent blank-turn tests.

Try the comparison before revealing the answer

Hide the EV columns and predict the choice for K♣ 9♣ on A♠ J♥ 6♦ 2♣, the same hand on Q♠ 7♥ 2♦ 3♣, and Q♦ J♠ on T♠ 7♥ 4♦ 2♣. Name a part of the checked range that matters. Then reveal all four action values and check whether your explanation matches the size of the gap.

For a hand without a pair or direct draw, compare checking before selecting a probe size, even after a low off-suit turn. Retained top pairs in the opponent’s range are a useful warning cue. The exceptions above still need their own hand and size comparison; neither an ace-high flop nor a paired turn decides the action by itself.

Our probe-betting guide covers the terminology, and the slow-playing casebook separates free-card risk from action value. In GTO Gecko, use available precomputed scenarios to compare actions, frequency distributions, EV estimates and ranges. The current app listing describes paid study plans and access by format.

Method and limits

We retrieved all 24 requested trees on September 12, 2026: eight flop trees and sixteen turn trees. The six held-out flop patterns contain 1,008 of the 1,388 eligible blank-turn hand rows. They were specified before EV analysis. There were no missing or excluded blank-turn boards.

Local loss is max(available action EVs) − chosen action EV. Within each board we use positive exported marginal combo reach; across the eight boards each receives equal weight. Turn ranges were checked against the flop action history and card removal, allowing for rounded weights and omitted near-zero combinations. Hero-specific response summaries also remove combinations conflicting with the hero’s cards.

Analyzed action vectors total 0.99, 1.00 or 1.01 because of the rounded export; we normalized them for frequency calculations. EVs are converted to bb using the current app’s scale. Three decimals reflect the file’s resolution, not verified solver accuracy; the 0.01bb near-tie tolerance is an editorial convention, not an error bound. The solver-convergence exercise explains why that distinction matters.

This is one stack depth, position pair, rake model and restricted size menu, with a forced BB flop check and unknown convergence. We did not evaluate a changed full policy, human error, ICM, multiway play or showdown equity. The selected low turns are not every card players might call a blank.

Download the selected hand values, flop check-back composition, response breakdowns and study method and arithmetic. Full proprietary trees stay private. Pio’s number definitions provide further context on range weights and matchup counts.