PulseGTO
Two flops with a king on top. Same preflop line, same source ranges before board removal, same tree. On K♠8♦3♣ the button's main play after a check is a quarter-pot bet. On K♠Q♦8♦ its most-used bet is a 125% overbet, while checking is its most frequent action overall. This article shows: - what the solver does on each flop; - which hands change their action; - what the range composition does and does not explain.
The setup
We re-solved both flops from the big blind's first decision, with one custom 6-max tree at 100 bb:
| Preflop | Button opens 2.5 bb, small blind folds, big blind calls |
| Flop | Pot 5.5 bb, 97.5 bb behind each player |
| Rake | 5 % of the pot, capped at 4.5 bb, no flop no drop |
| Big blind | Checks, or leads 50 % |
| Button after a check | Checks, or bets 25 % / 50 % / 75 % / 125 % (1.38 / 2.75 / 4.12 / 6.88 bb) |
| Raises | 3x on flop and turn, 2.5x on river, all-in allowed |
| Turn and river | One bet size: 75 % |
| Accuracy | K83 0.29 %, KQ8 0.25 % of the pot (PulseGTO RangeCFR) |
The ranges are the ones this custom tree reaches on this line. They are not universal charts. The same tree is used for both flops. Board removal and the solved BB checking strategy affect the ranges at the decision shown.
Result: the contrast holds
The button's strategy after the big blind checks:
| board | check | 25 % | 50 % | 75 % | 125 % |
|---|---|---|---|---|---|
| K♠8♦3♣ | 34.7 % | 41.3 % | 19.7 % | 3.2 % | 1.0 % |
| K♠Q♦8♦ | 56.4 % | 3.6 % | 5.8 % | 6.6 % | 27.5 % |
- K83: quarter pot is the most used action. Half pot adds another 20 %, and the overbet is almost unused.
- KQ8: the overbet is the main bet, and the button checks more than half its range. The button bets less often in total (43.6 % against 65.3 %), and with a larger size.
Bet amounts are rounded to 0.01 bb. These are finite-accuracy re-solves; the displayed frequencies describe this model, rather than exact universal GTO.
Follow the same hands
The clearest way to see the difference is to follow a few hands across both boards. The numbers are averages over all combos of each hand, except J♦T♦, which is one combo.
| hand | K♠8♦3♣ | K♠Q♦8♦ |
|---|---|---|
| 88, set | 25 %: 49.7 · 50 %: 37.8 · 75 %: 9.3 | 125 %: 91.7 · 75 %: 7.6 |
| AKo, top pair top kicker | 25 %: 42.4 · check: 33.6 · 50 %: 21.0 | 125 %: 60.2 · check: 24.2 · 75 %: 9.9 |
| 99, pair below the top card | 25 %: 60.2 · 50 %: 26.1 · check: 9.7 | check: 97.0 · 25 %: 3.0 |
| J♦T♦ | 50 %: 52.2 · 25 %: 37.8 (backdoor flush draw) | 125 %: 76.9 (open-ended straight draw + flush draw) |
- The strongest hands and good draws move from small and medium bets on K83 to the overbet on KQ8.
- A middle pair like 99 does the opposite: it bets on K83 and checks on KQ8.
By hand class on KQ8: - Overbet most: sets 80.8 %, overpairs 73.7 %, two pair 58.4 %, top pair with a flush draw 53.3 %. - Draws: gutshots, flush draws and open-enders overbet 40–42 % of the time. - Check: pocket pairs below the top card 96.9 %, second and weaker pairs 81.3 %. - Top pair as a whole checks 57.8 %; the best kickers overbet.
On K83 every hand class uses the quarter-pot bet between 34 % and 55 % of the time.
What the main bets are made of
| action | two pair or better | overpair | top pair | weaker pair | draw, no pair | no pair, no draw |
|---|---|---|---|---|---|---|
| K83: bet 25 % | 5 % | 1 % | 15 % | 28 % | 19 % | 33 % |
| KQ8: bet 125 % | 15 % | 3 % | 13 % | 14 % | 36 % | 19 % |
| KQ8: check | 1 % | 0 % | 13 % | 47 % | 16 % | 22 % |
- KQ8 overbet: built mostly from top pair or better (31 %) and from draws without a pair (36 %).
- KQ8 checks: almost half are weaker pairs.
- K83 quarter-pot bet: spread across every kind of hand.
How the big blind answers
| big blind facing | fold | call | raise |
|---|---|---|---|
| K83, 25 % bet | 29.1 % | 51.1 % | 19.8 % |
| K83, 50 % bet | 40.7 % | 50.0 % | 9.3 % |
| KQ8, 125 % bet | 61.6 % | 33.4 % | 5.0 % |
The small bet on K83 is not free: the big blind raises it almost one time in five.
What might explain the switch?
The saved outputs show different range compositions: on KQ8 there are more pairs and draws, and the overbet mixes strong made hands with draws. On K83 the smaller bet is spread across more hand categories. That is a useful description of the strategies, not an isolated test of their causes.
One hypothesis is that the different distribution of medium-strength hands and draws changes what each sizing achieves. But a fixed equity threshold, such as “hands with at least 80% equity,” is not a standalone measure of nut advantage. It does not capture all future runouts, blockers or how the strongest parts of the ranges interact.
The accompanying video discusses this interpretation. Treat it as a hypothesis to investigate, rather than proof that one feature explains the overbet. Comparing individual hands gives us the clearest supported observation: sets and strong draws often use the large bet on KQ8, while hands such as 99 check.
Limits
- Scope: two flops, one custom tree, one set of preflop ranges, one rake structure.
- Simplified tree: turn and river have only one bet size.
- Accuracy: about 0.25–0.29 % of the pot. Small frequency differences are not meaningful, and decimals will move with further solving.
- Not measured: how much EV a simpler strategy would lose (for example, one bet size for the whole range). That needs a separate restricted solve.
- Coarse labels: the hand-class buckets are coarse, and equity is the solver export's equity against the other range at the decision. It is not a full study of equity realisation.
Takeaway
When two boards look alike, don't stop at the favourite bet size. Follow a few hands, such as a set, top pair and a middle pair, and see which of them change their job. Here the set and the best draws switch to the overbet on KQ8, while pocket nines switch from betting to checking.
Analysis made with PulseGTO on a custom 6-max 100 bb tree. More studies: pulsegto.com/blog.