22 Bet-Sizing & Timing as Information

In the previous chapter you learned to build and narrow an opponent’s range from the actions they take: who raised, who called, what the board did, how the story holds together. This chapter is about a second, parallel stream of information that runs alongside the what of an action — the how. Two players can both bet the river. One slides out a third of the pot; the other shoves two pots in. Those are not the same bet. One snap-calls in a quarter second; the other tanks for forty. Those are not the same call.

Bet sizing and timing are signals. Against unbalanced opponents — which is to say, against the large majority of the player pool below the highest stakes — these signals leak a remarkable amount of information about hand strength. The job of this chapter is to teach you to read them, to fold them into the range read you built in Chapter 21, and to do so without falling into the trap that catches most aspiring hand-readers: treating a probabilistic lean as a certainty, and walking into the trap a thinking opponent has laid precisely because they know you are watching.

Key idea

Sizing and timing are not a separate skill from hand reading — they are inputs to it. A read is never “the small bet means weak.” A read is “given everything I already knew about this player’s range here, the small bet shifts the weight of that range toward the weak/blocker/thin-value region.” You are updating a distribution, not flipping a switch.

22.1 Why sizing leaks information at all

A theoretically perfect player chooses bet sizes so that, at every size they use, their range is balanced — the ratio of value hands to bluffs is calibrated so you cannot profitably exploit them no matter what you do. Crucially, such a player uses each size with both strong and weak hands, so the size itself tells you nothing about which they hold.

Almost nobody does this. Most players instead let the size follow the hand: they pick a number that “feels right” for the holding in front of them. Strong hands want to get paid, weak hands want to get there cheaply or steal cheaply, and the bet size becomes a transcript of that internal feeling. The entire edge in this chapter comes from one fact: for unbalanced players, the size is chosen as a function of their actual hand, so it betrays the hand.

This is why the same physical bet means different things from different players. A 75% pot turn bet from a balanced regular is close to noise. The identical bet from a recreational player who only ever bets that big with two pair or better is a flashing sign. You are not reading the bet; you are reading the bet relative to that specific player’s tendencies. Build that baseline first — from showdowns, from the HUD if you are online, from the last two hours at the table if you are live — and read the deviation from it.

22.2 A vocabulary of sizes

Let me lay out the common sizing buckets and what they typically encode from a “size follows hand” opponent. Treat every claim here as a population lean, not a law.

Small bets (roughly 20–40% pot)

Small bets are the most polysemous size in poker, and you must use the rest of the range read to disambiguate. From unbalanced players a small bet usually means one of:

Genuine weakness asking for a cheap continuation — a marginal made hand or a weak draw that wants to see a card or take it down without committing. Common in single-raised pots when the bettor has, say, second pair and no plan.
A blocker / “merge” bet — a hand good enough to bet thinly but not strong enough to want a raise. Online, this overlaps heavily with what solvers actually do (small c-bets on dry, range-advantaged boards), so against competent players a small flop bet is not a tell at all; it is correct strategy.
Thin value milking a call — a player who has a good-but-not-great hand and sizes down because they only get called by worse if the price is small.

The single most important live-versus-online distinction: on dry, static boards where the preflop aggressor holds a range advantage (think A K J K-7-2 rainbow as the raiser), a small flop bet is standard, correct, balanced play. Do not “read” a small c-bet there as weakness — you will be reading a solver output as a tell. The small-bet-equals-weak read applies most reliably to turns and rivers, and to out-of-position bets and donk bets from players who have no business knowing what a merge bet is.

Pot-ish bets (roughly 60–100% pot)

This is the “default” zone, and by itself it is the least informative size because so many hands pass through it. Where it does talk is in specific player types and spots:

In recreational and lower-stakes live games, a sudden jump to a near-pot or pot-sized bet on the turn or river — from a player who has been betting half-pot all night — is frequently raw strength. They finally have a hand worth “betting it like they mean it.” The deviation is the signal.
A pot-sized bet that commits a short stack is often a player who has decided to get it in and is choosing the size that does so, i.e., value or a committed draw.

Large bets and overbets (110% pot and up)

A bet larger than the pot is, by its nature, a polarizing size: mathematically it only makes sense with a range of strong value hands and bluffs, because you are risking a lot to win comparatively little, which requires either the best hand often or fold equity. From unbalanced players the read splits cleanly by player type:

The straightforward / “honest” recreational player who overbets is almost never bluffing. Their overbet means the near-nuts. They have a monster and they are afraid you will fold, so — paradoxically — they bet huge because in their mind “I want to win a big pot.” Against this profile, a river overbet is one of the most reliable strength tells in poker, and the correct response is to fold everything but your own strongest hands and a few specific blockers.
The aggressive, “fancy” regular or a known maniac who overbets is doing it because they know the size looks polarized and represents the nuts — which is exactly why they will use it as a bluff. Against this profile the overbet is far closer to balanced, and you must defend wider, leaning on blockers and on the rest of the range read rather than on the size.

This split is the heart of the chapter’s recurring warning: the very same action carries opposite information depending on who performs it. You cannot read sizing without first classifying the player.

Common mistake

Applying a sizing read across player types as if it were a law of physics. “Overbet means nuts” is true for the honest recreational player and dangerously false for the thinking regular who knows you believe it. Before you act on a sizing read, ask: does this player know what this size is supposed to mean? If yes, downgrade the read — and consider that they may be weaponizing it.

The “wrong size for the hand they’re repping” tell

A higher-order read, available even against semi-competent players: when the size does not match the hand the story is supposed to be telling, something is off. A player who flats the flop and then, on a blank turn, suddenly fires a huge overbet is often over-betting because they are afraid of the hand they hold — either a vulnerable made hand that wants you out, or a bluff that needs maximum fold equity. Conversely, a tiny “please call” river bet into a scary board frequently means a value hand begging for a crying call. Size that screams urgency usually is urgency, in one direction or the other; your job is to use blockers and the range read to decide which.

22.3 Timing tells

Timing is the second channel. The governing principle is intuitive: the brain spends time on hard decisions and little time on easy ones. A hand that is obviously a bet takes no thought; a hand that is genuinely difficult takes time. From there, the standard reads:

The snap

A snap-bet (instant bet) usually means the action was pre-decided, which most often means a planned bet — either a routine value bet or, importantly, a planned bluff on a card the player had already decided to barrel. Snap-bets are therefore less reliably strong than people think; what they reliably mean is “no decision was required,” which can be a clear value hand or a pre-loaded bluff.
A snap-call typically means a hand that beats some bets but isn’t strong enough to raise and isn’t weak enough to consider folding — a comfortable medium-strength catch. Critically, a snap-call often denies the caller the strongest hands: with the nuts they would at least consider raising, and that consideration takes time. So a snap-call frequently caps the range below the top. It also denies the bluff-catching tank — they didn’t agonize, so they weren’t close to folding.
A snap-check generally means giving up or a hand with no intention of building a pot — though against tricky players it can be a pre-planned check-raise, which is the timing version of the overbet trap.

The tank

A tank-then-call is one of the more reliable live tells: genuine deliberation usually means a genuinely marginal hand. Players rarely tank for a long time and then call with the nuts (they’d be deciding whether to raise, and even then a slowroll is socially costly), and they rarely tank long with pure air they intend to fold. Tank-call therefore leans toward a mid-strength bluff-catcher — exactly the hand they were unsure whether to fold.
A tank-then-raise or tank-then-shove is the dangerous one. In live play, a long tank followed by aggression is, more often than not, a strong hand performing reluctance — and “Hollywooding” (acting weak when strong) is a real, documented pattern in live poker. The old adage strong means weak, weak means strong captures the deliberate-acting case: a player who sighs, shrugs, and then raises is frequently holding a monster and managing your perception. This is not universal, but the base rate of the long-tank-into-raise being value is high enough that you should rarely hero-call it without a specific reason.
A tank-then-fold is just a close fold and tells you mostly about their range on future streets (they were near the threshold).

Key idea

Decompose every timing read into two questions. First: was a decision required? (snap = no, tank = yes). Second, only for genuine reads: which direction does the difficulty point? A genuine tank means a genuinely marginal spot; a deliberate, theatrical tank-into-aggression usually means the opposite of what it performs. Distinguishing genuine deliberation from performance is the live player’s core timing skill, and it is read from the whole person — posture, breathing, where the chips come from — not the clock alone.

Online timing

Online strips away the body but adds a clean, quantifiable clock, and that clock is often more reliable than live tells because it cannot be faked with a poker face — only with deliberate effort that most players don’t bother to apply.

Instant action via pre-action buttons (“bet/raise” or “check/fold” boxes that fire the moment it’s your turn) is the strongest online timing tell. An instant check very often means the check/fold box was ticked — the player has given up. An instant call can mean the check/call box was pre-clicked, indicating a hand they had already decided was a call regardless of the card. Learn what the instant-action artifact looks like on your client; it is close to a free read.
Normal-tempo actions (a second or two) are mostly noise — the player is just clicking in rhythm.
Long pauses that use the time bank mean a genuinely hard decision. As live, a tank-into-raise online leans strong; a tank-into-call leans marginal. But beware: a delay can also be multi-tabling lag — the player was acting on another table, or got distracted — which is pure noise, not a read. The more tables someone plays, the more their timing degenerates into noise and the less you should trust any single delay.
Consistent, robotic timing across every decision is itself information: it usually marks a strong, multi-tabling regular or someone using a tool/script to standardize tempo precisely so as to not leak. Against robotic timing, abandon timing reads and lean entirely on sizing and range.

Common mistake

Trusting a single online delay as a tank tell while multi-tabling is rampant. A four-second pause from a known 12-tabler is probably them clicking back from another table, not deliberation about your bet. Calibrate the player’s baseline tempo first; only deviations from their own normal mean anything.

22.4 Combining sizing + timing + the range read

None of these signals should be used alone. The method is to take the range you built in Chapter 21 and let sizing and timing re-weight it. Think of it as three filters stacked on top of one hand-distribution:

Range read (ch. 21): From the line, the board, and preflop ranges, what hands can they have here, and with what rough weights?
Sizing filter: Given this player’s size-follows-hand tendencies, which part of that range does this size emphasize? Polarize, depolarize, or cap.
Timing filter: Does the tempo confirm or contradict the sizing read? Concordant signals compound your confidence; contradictory signals should lower it, not be cherry-picked.

The most powerful situations are when sizing and timing agree and point at a narrow slice of the range. The most dangerous are when they disagree — that is precisely when a thinking opponent may be manufacturing a false tell, and your correct response is humility: revert toward the unexploited range read and don’t stack off on a single signal.

Drill

For your next three live or online sessions, before you act, say to yourself in one sentence: “Range says ; the size shifts it toward ; the timing ___ that.” Write down five hands where you reached showdown and check whether the size/timing read was correct. You are building a personal hit-rate on these reads — which is the only honest way to know how much to trust them. Expect to be wrong a meaningful fraction of the time; that is normal and is why these are weights, not certainties.

22.5 Worked example

Live $2/$5, 100bb effective. You are on the button with K Q. A loose-passive recreational player (showdown history: bets small with marginal hands, jumps to big bets only with two pair or better, has never been caught bluffing the river) limps UTG. You raise to 25 (5bb), and the limper calls. Heads-up to the flop.

Flop (≈ 11bb pot): Q 9 4, two hearts. The limper checks, you bet 7 (about 60% pot) for value and protection with top pair good kicker, and he calls fairly quickly. His range here is wide: pairs (queens with a worse kicker, nines, fours sets), flush draws, straight-draws like J-10, and floats. Nothing surprising.

Turn (≈ 25bb pot): 4. Board Q 9 4 4. He leads — donks out — for 20, about 80% pot, after a short pause. Now stack the filters:

Range read: A donk-lead from a passive player is already a deviation from his normal “check to the raiser” behavior. The board paired the bottom card. Hands that improve to trips on this exact card are 4x, and the spades/clubs draws didn’t get there. His donk credibly represents a four, a slowplayed nine or set from the flop now feeling urgent, or — possibly — a flush draw turning into a bluff, though this player “has never been caught bluffing.”
Sizing filter: 80% pot is, for this player, a “big bet,” and his baseline is that big bets mean two pair or better. The size strongly emphasizes the trips/full-house region (any four, especially 9-4/Q-4, or a flopped set now boated). It de-emphasizes the bluff region, because this profile does not fire big as a bluff.
Timing filter: The pause-then-lead is mild deliberation — consistent with a player who flopped or just made a hand he’s now deciding how to play, summoning the courage to lead. It is not a snap (not pre-loaded) and not a long Hollywood tank-into-raise. It modestly confirms “I have something and I want to bet it,” which aligns with the sizing read.

All three filters point the same direction: this is very likely a four or better. Your K Q is a single pair. Against this specific, honest, non-bluffing player whose big-bet-donk on a paired board screams trips, the disciplined fold is the high-EV play. You fold, and you do it without agonizing, because the reads are concordant and the player type is one whose tells are reliable.

Now change one variable. Suppose the donk-leader were instead a tough, aggressive regular known to attack paired boards as bluffs and to use the donk-lead as a designed line. The identical 80%-pot lead now means something close to the opposite: a paired board is exactly where a thinking player bluffs, because your range is full of one-pair hands that can’t continue. The sizing tell inverts, the range stays bluff-heavy, and against that opponent K Q — top pair, a heart blocker reducing some of his flush-draw-turned-bluff combos — becomes a defensible call. Same board, same bet, same timing; opposite conclusion, entirely because of who is betting. That contrast is the whole chapter in one hand.

22.6 A short discipline for using tells

To keep these reads from becoming superstition, hold yourself to a few rules:

Establish the baseline before you read the deviation. A tell is a departure from a player’s own normal sizing and tempo. No baseline, no tell.
Classify the player first. “Does this person know what this size/timing is supposed to mean?” Honest players leak; thinking players weaponize. The read flips on that one question.
Require concordance for big decisions. Stack off on sizing + timing + range all pointing the same way. When signals conflict, shrink your confidence and default to the GTO-ish range read.
Track your hit-rate. You will be wrong often — these are probabilistic edges, not x-ray vision. Logging showdowns is the only way to calibrate how much weight a given read deserves, and to catch yourself seeing patterns that aren’t there.
Assume strong players are reading you too. The moment your reactions become predictable, your own sizing and timing become a tell for them. Vary your tempo, standardize your sizes against good opponents, and never let your bet size narrate your hand to someone capable of listening.

Key idea

The exploitative edge in sizing and timing comes almost entirely from opponents who let their action follow their hand. Harvest that edge ruthlessly against the unbalanced majority — and, against the players who don’t leak, fall back on the balanced range reading of the previous chapter and refuse to be tricked by a tell that was placed there for you to find.