Texture switches sides on the river

Triple-barrel% (aggressor, after bet-bet) vs fold% (defender, facing the third barrel) · single-raised pots BTN vs BB

Aggressor: barrel%Defender: fold%
0%10%20%30%40%50%60%70%Trips on BoardTrips on Board · barrel 45.8%Trips on Board · fold 41.0%46 / 41%Two Pair on BoardTwo Pair on Board · barrel 37.9%Two Pair on Board · fold 46.5%38 / 47%Second Card PairsSecond Card Pairs · barrel 39.4%Second Card Pairs · fold 46.7%39 / 47%Third Card PairsThird Card Pairs · barrel 47.0%Third Card Pairs · fold 49.5%47 / 50%Top Card PairsTop Card Pairs · barrel 47.6%Top Card Pairs · fold 51.1%48 / 51%Ace OvercardAce Overcard · barrel 55.8%Ace Overcard · fold 52.4%56 / 52%Bottom Card PairsBottom Card Pairs · barrel 51.6%Bottom Card Pairs · fold 53.5%52 / 54%Undercard + StraightUndercard + Straight · barrel 50.9%Undercard + Straight · fold 54.1%51 / 54%BlankBlank · barrel 54.7%Blank · fold 55.6%55 / 56%Broadway OvercardBroadway Overcard · barrel 59.6%Broadway Overcard · fold 56.5%60 / 57%Flush + UndercardFlush + Undercard · barrel 47.0%Flush + Undercard · fold 56.8%47 / 57%Overcard + StraightOvercard + Straight · barrel 56.5%Overcard + Straight · fold 58.8%57 / 59%Flush + OvercardFlush + Overcard · barrel 50.2%Flush + Overcard · fold 60.1%50 / 60%Mid OvercardMid Overcard · barrel 54.8%Mid Overcard · fold 60.1%55 / 60%4-Straight4-Straight · barrel 44.1%4-Straight · fold 61.1%44 / 61%Triple BroadwayTriple Broadway · barrel 53.6%Triple Broadway · fold 62.4%54 / 62%4-Flush4-Flush · barrel 44.8%4-Flush · fold 67.1%45 / 67%
Sorted by fold%. The attack bars barely move — a player who bet flop and turn fires most rivers at 38–60% regardless of the card. The defense bars climb 26 points, from trips on the board (41% fold) to the fourth flush card (67%). On the river, texture stops being a bluffing map and becomes a folding map.
In one minute
  • The third barrel is nearly texture-blind. After betting flop and turn, the population fires at a similar clip on almost every river — a fraction of the spread the same cards produce on the flop.
  • The fold side is where texture lives now. Folding to the third barrel swings 26 points by river class — the widest defensive texture response of any street.
  • Pairing rivers are calling rivers. Trips on the board is the least-folded river in poker (41%): nobody folds a hand the whole table shares.
  • Scare cards actually scare. A fourth flush card folds out 67%; a board that turns all-broadway, 62%.
  • Two pair on the board is the river to raise. Lowest barrel (38%) and the highest check-raise rate on the street (10%).

The experiment

We used the same method as our flop and turn studies: measure what regulars actually do on every possible card, hand a decision tree the raw ingredients — rank relation to the board, suit transitions, straight windows, pairing — and keep only the boundaries that replicate on independent halves of the player pool. The spot: the aggressor bet flop and turn and got called twice, and now decides on the river — 146,000 triple-barrel decisions — while on the other side the defender, having check-called twice, faces the third bullet (44,000 decisions).

The axis that vanishes

On the flop, texture explains most of what the population does (the board’s gap structure). On the turn, less, but still plenty (the card’s relation to the board). On the river, it nearly disappears: the best class map we can build explains a fraction of what it does one street earlier. It makes sense once you say it out loud — a player who has already bet twice has committed to a story; by the river his decision is driven by his own hand and the pot odds he is laying, not by the card that just landed.

The axis that reappears — on the other side

The defender, who has check-called twice, is the one actually listening to the river. His fold-vs-barrel by class is the steepest texture ladder we found on any street, on either side:

Fold vs the third barrel, by river class

BB after check-calling flop and turn, single-raised pots · 44k decisions

0%10%20%30%40%50%60%70%4-Flush4-Flush · 67.1% fold67.1%Triple BroadwayTriple Broadway · 62.4% fold62.4%4-Straight4-Straight · 61.1% fold61.1%Flush + OvercardFlush + Overcard · 60.1% fold60.1%Mid OvercardMid Overcard · 60.1% fold60.1%Overcard + StraightOvercard + Straight · 58.8% fold58.8%Flush + UndercardFlush + Undercard · 56.8% fold56.8%Broadway OvercardBroadway Overcard · 56.5% fold56.5%BlankBlank · 55.6% fold55.6%Undercard + StraightUndercard + Straight · 54.1% fold54.1%Bottom Card PairsBottom Card Pairs · 53.5% fold53.5%Ace OvercardAce Overcard · 52.4% fold52.4%Top Card PairsTop Card Pairs · 51.1% fold51.1%Third Card PairsThird Card Pairs · 49.5% fold49.5%Second Card PairsSecond Card Pairs · 46.7% fold46.7%Two Pair on BoardTwo Pair on Board · 46.5% fold46.5%Trips on BoardTrips on Board · 41.0% fold41.0%
Pairing rivers get looked up; completions get believed. The two Four Card+ classes fold out 61–67% — a third barrel on those cards prints. And the population knows it from the other side too: check-raises peak on flush-completing rivers and on two pair on the board, where the defender’s own range just got there.

Which card pairs, one street later

Our turn study found that pairing the board’s middle card is among the most bet-hostile cards in the deck. The river says the same thing in its own vocabulary. Pairing the second board card collapses barreling to 39.4% and draws one of the lowest fold rates of the street (46.7%) — the cards a caller peels with live in the upper-middle of the board, so pairing there is where his range improved and yours did not. Pairing the bottom card barely registers (51.6%). And on a board that was already paired, a river that pairs a second rank puts two pair on the board: 37.9% barrel, and the highest check-raise rate on the street (10.0%) — everyone holds the same two pair, so the only way to win the pot is to take it. Its mirror, trips on the board, is the least-folded river in poker at 41.0%.

The map

Seventeen river classes, mirroring the turn map

Barrel% (single-raised BTN vs BB IP) · fold% and check-raise% (BB vs the third barrel)

ClassDefinitionExampleShareBarrelFoldX/R
Overcard — clean: no flush, no straight arrives
Broadway OvercardK, Q or J above all four cardsQ on T♦7♠4♥2♣3.9%59.656.56.3
Ace Overcardan ace arrivesA on T♦7♠4♥2♣1.8%55.852.45.8
Mid Overcard7–T, still above the board9 on 6♦4♠3♥2♣0.9%54.860.14.9
Blank
Blanknothing lands at all4 on K♦9♠6♥2♣22.2%54.755.65.3
Flush Completes
Flush + Overcardovercard that completes a flushQ♠ on 8♠5♠2♥K♦2.9%50.260.110.1
Flush + Undercardcard under the highest board card that completes a flush7♠ on K♠9♠6♥2♣14.3%47.056.89.0
Straight Completes
Triple Broadwaythird broadway card lands (any of them)J on K♦T♠7♥2♣7.0%53.662.44.5
Overcard + Straightovercard that completes a straight9 on 7♦6♠4♥2♣1.9%56.558.86.0
Undercard + Straightcard under the highest board card that completes a straight8 on T♦7♠2♥K♣15.5%50.954.15.7
River Pairs
Top Card Pairspairs the highest board cardK on K9723.3%47.651.16.8
Second Card Pairspairs the second-highest9 on K9723.4%39.446.77.9
Third Card Pairspairs the third-highest7 on K9723.8%47.049.58.2
Bottom Card Pairspairs the lowest board card2 on K9724.0%51.653.55.2
Two Pair on Boardpairs a second rank — the board is now double-paired7 on K9972.7%37.946.510.0
Trips on Boardpairs the already-paired rank9 on K9971.0%45.841.07.8
Four Card+
4-Straightputs four (or five) to a straight on the board6 on 7♦5♠4♥K♣8.4%44.161.16.8
4-Flushfourth (or fifth) card of a suit2♠ on K♠9♠6♠4♥3.0%44.867.16.4
The same blueprint as the turn map, minus Double Flush Draw (no next street to draw to) and with the pair ladder at four positions plus the two board-repair classes. Every card gets exactly one class. Resolution order: 4-Flush, 4-Straight, the flush completions, the Straight Completes family, then clean overcards and pairs. Five-card events are absorbed upward: the fifth suited card (0.1% of rivers) counts as 4-Flush, a straight landing on the board (0.2%) as 4-Straight — both far too rare to earn classes, though the 5-flush is quietly the most folded-to card in poker (71%).

The scorecard

On the same taxonomy-agnostic grid of river situations, a traditional 17-class river map explains 0.063 of class-explainable barrel variation (split-half: 0.044), 0.046 of the folding and almost none of the raising. The seventeen classes above score 0.163 on the barrel (split-half 0.109), 0.119 on the fold and 0.065 on the raise — double to sixfold, from the same number of classes. Both barrel numbers stay small, and that is the finding: the river card does not move the bettor much. It moves the caller — and for the caller, this map more than doubles what the traditional one sees.

Method note. Sample: same corpus as the flop and turn studies — 94.6M 6-max cash hands, regulars only, NL50–NL1000. Node: aggressor bet flop and turn, got called twice, faces the river root decision, in three configurations (3-bet pot OOP, 3-bet pot IP, single-raised IP); defense = BB check-called flop and turn and faces the third barrel (fold / call / check-raise). Map construction identical to the earlier studies: CART over raw card features, minimum class share 2%, boundaries validated on held-out halves of the player pool. Scores are reported on a taxonomy-agnostic grid of river situations; absolute values depend on that grid, so only same-grid comparisons are meaningful. No solver output was used anywhere in this article.