The 7 Pillars of Poker Strategy
and what our solver says about each

Every poker player eventually realizes that when to bet matters less than how much. Bet sizing is where edges compound — and where most players get it wrong, not because the right size is hard to find, but because the pattern is counter-intuitive.

This pillar catalogues six sizing concepts from modern GTO poker theory and tests each against our solver at scale. Every number on this page comes from running real queries against QuintAce's own model — not from quoted literature. When we say the solver bets K72 rainbow 83.6% of the time for 2.4bb, that's a query we ran, a result we recorded, and a finding we verified for consistency across multiple research rounds.

Why small bets on dry boards, big bets on wet ones — and small again on very wet ones

Most players know to bet small on dry boards and bigger on wet ones. Fewer realize the relationship isn't monotonic — it's a parabola. Past a certain wetness, sizing drops again.

Here's the solver's c-bet strategy across 12 canonical flop textures. Same position (CO vs BB), same stack depth (100bb), same preflop action. Only the flop changes.

CO c-bet frequency and average size on 12 flop textures, 100bb single-raised pot.

Source: cash-baselines.md lines 65–80

Dry K-high boards get the highest frequency at ~85% with medium sizes. Wet connected boards drop to ~55% frequency but hold their size. Monotone boards see a drastic drop on both axes — 32% frequency on K94ss, the smallest sizes in the table. More wetness does not mean more betting.

Want to see this in your own hands? Drop any flop into QuintAce's solver and the range analysis shows you how sizing changes across the wetness spectrum in 10 seconds.

You can't bet big unless two things are true

Nut advantage and fold equity are the two primary drivers of bet sizing. Not one or the other — both, together, at the same time.

The solver goes big when both of these are true:

1. You have more strong hands than your opponent (nut advantage)
2. Your opponent has hands that will actually fold (fold equity)

Miss either one and big bets stop working. When we tested this at scale in Round 9, the pattern showed up on every board category we checked:

• KK5 (paired, no clean nut advantage) — 99% small bets. Neither side holds the nuts clearly, so there's no foundation for large sizing.
• K94ss (monotone, nut advantage but no fold equity) — 90% small bets. The raiser's range still has value, but the opponent's flush draws won't fold. Condition 1 holds, condition 2 fails.
• T98 (wet connected, both conditions partial) — 71% medium. Some value, some fold equity, medium sizing is the middle ground.
• K72r (dry K-high, both conditions strong) — 62% medium + 38% small. Both line up cleanly. The solver still splits the sizing — medium for hands that want pot growth, small for hands that want to see a cheap turn.

Source: cash-baselines.md line 212 (R9 verdict notes, theory D3)

The two-condition rule explains the wetness parabola directly. Dry boards pass both. Wet boards start failing on fold equity. Very wet boards fail fold equity entirely. Paired boards fail nut advantage. Every point on the curve is a point where zero, one, or two conditions hold.

QuintAce's solver view shows range vs range equity on any flop in seconds — including which side has the nut advantage and which part of the opponent's range will actually fold.

Overbets on the flop are vanishingly rare — here's why

If big bets need two conditions, overbets (more than 100% of pot) need a third: the opponent's range must be capped — prior actions must have removed their strongest hands.

On the flop, both players still have their full preflop ranges. Neither has had the chance to signal weakness by checking. So the "capped opponent" condition almost never holds. When we checked in Round 9, the model used overbets on just 0.0–0.3% of flop bets across all seven boards tested. Flop overbets aren't wrong — they're mostly unnecessary.

Overbets belong to later streets, after someone checks. A check signals "I don't have a strong value range right now," which caps the checker's range and opens the door for the non-checker to overbet on the next street. It's a turn-and-river tool, not a flop one.

Source: cash-baselines.md line 213 (R9 verdict notes, theory D4)

Bets grow geometrically across streets

Zoom out from the flop. Over a full multi-street line, bet sizes grow geometrically — each street pushes the pot toward a target (often all-in), with each bet a roughly constant fraction of the pot at that moment.

When we ran K72r through the solver across three streets with blank turns and rivers, here's what the average bet sizes looked like:

Average solver bet size across a three-street line on two test boards — K72 rainbow and A94 rainbow — CO vs BB, 100bb single-raised pot. Both boards show the same ~2.5–3× per-street growth pattern.

Source: cash-baselines.md line 226 (R10 verdict notes, theory D1)

Bets grow by roughly 2.5× to 3× per street — close to the geometric ideal where each bet leaves the right amount for the next one. On A94r the pattern repeats: flop 2.35bb, turn 8.29bb, river 20.6bb.

Frequency drops hard alongside sizing growth. The solver isn't barreling the same hands three times for increasing amounts — it's tightening the range each street. Most weak bluffs give up on the turn. The remaining value concentrates into the river. By the time the river arrives, the betting range is small enough to support huge bets.

This theory sits at moderate evidence because the "geometric" framing is partially our own — we arrived at it by combining the solver's multi-street behavior with pot-geometry math. The direction and magnitude are solver-verified.

Shallow stacks compress bet sizes and widen betting frequency

At 20 big blinds, the same decisions look nothing like they do at 100. Bets are smaller. Betting happens more often. Strategy feels flatter — less polarized, more utilitarian.

When we extended the test in Round 12 across six stack depths on K72r:

Average CO c-bet size on K72 rainbow across stack depths, CO vs BB single-raised pot.

Source: cash-baselines.md line 199 (R12 verdict notes, theory D5)

Sizing monotonically grows with stack depth — a 20bb stack produces a 25% smaller bet than a 100bb stack on the same flop.

The frequency side turned out more nuanced than the original theory. Frequency peaks at 50bb (89.6%), then declines in both directions — down to 75.7% at 200bb. The theory said "shallow stacks bet more often." The data says "shallow-to-medium stacks bet more often; very deep stacks bet less often." That's a useful refinement, and exactly the kind of thing verifying theories against real solver output is supposed to catch.

River sizing splits by blockers — the same hand can take two sizes

The most subtle sizing finding in modern theory: on the river, the same hand category can take two different bet sizes based on blocker cards.

A set of kings blocks the combos of top pair kings. If the opponent would have folded top pair anyway, blocking those combos is bad — it shrinks the opponent's calling range. So top set should bet smaller on a K-paired river. Bottom set, which doesn't block top pair, can bet larger and get called more often.

That's the theory. When we ran the test in Round 10 on K72r → 2d → Kc (a K-paired river):

• KK splits between bet12.8 (18% of combos) and bet17.1 (81% of combos)
• QQ splits between bet8.6 (74%) and bet12.8 (25%)

The same strong hand category takes two different bet sizes. Size splitting is there, and it's clean.

Source: cash-baselines.md lines 223–224 (R10 verdict notes, theory D6)

What we couldn't confirm is which specific blocker cards drive the split. Isolating that would need combo-level extraction — asking "for this exact KcKh vs this exact KsKh, which size does the solver prefer?" Our query infrastructure gives hand-class splits, not combo-level blocker isolation. The existence of blocker-driven splitting is verified; the specific mechanism is still under-isolated. We flag this as partial.

QuintAce's hand analysis view shows the full combo-by-combo breakdown on any river spot — including the mixed-sizing strategies the solver actually uses.