21 The Hand-Reading Method: Range Narrowing Street by Street

By the end of this chapter you will be able to narrow an opponent’s range street by street, from their preflop opening range down to a small, weighted river range you can confidently fold, call, or raise against.

An earlier chapter argued that you must think in ranges rather than single hands. That was the concept. This chapter is the procedure: the repeatable, almost mechanical process you run hand after hand to turn “I have no idea what he has” into “his river range is roughly these twelve combinations, weighted toward value, and against that I fold.” Hand reading is not a psychic gift reserved for a few naturals. It is a disciplined funnel anyone can learn, and like any skill it gets faster and more accurate with reps until it runs in the background.

The whole method fits in one sentence: start with the widest range your opponent could have given their position and preflop action, then on every street remove the hands that would not have taken the action they actually took, paying attention to weights, until you reach the river with a small, lopsided range you can make a decision against.

Everything below unpacks that sentence and then walks a single hand through the entire funnel.

21.1 The Funnel: A Five-Step Loop

The process repeats on each street with the same four questions, bracketed by a setup step and a payoff step:

Set the prior. Before a single community card, assign the opponent a starting range from position and preflop action.
Observe the action. They check, bet, raise, or call, at a particular size and a particular speed.
Ask the filter question. “Of all the hands still in their range, which ones would actually take this action, this size, on this board?”
Update. Remove the hands that fail the filter; lower or raise the weight on the rest. The survivors become your new, narrower range.
Repeat on the next street. The narrowed range from step 4 is the prior for the next card. At the river, stop and use the final weighted range to make your decision.

That is the entire engine. The art is in step 3: answering the filter question honestly, for this specific human, rather than for an imaginary “balanced” robot. The rest of this chapter teaches you to answer it well.

Key idea

You are never trying to name your opponent’s two cards. You are trying to describe the shape of their range, how much of it is value, how much is bluff, how much is give-up, because that shape, not any single guess, is what your call, fold, or raise plays against.

21.2 Bayesian Updating in Plain Language

The funnel is just Bayes’ theorem applied to poker. You do not need the formula, but you should understand the three ideas under it, because they prevent the two most common hand-reading errors.

The prior is the range you start with before the new action: the distribution of hands the opponent could hold. At the top of the hand this is their positional opening range; after the flop it is whatever range survived the flop.
The likelihood is the key question: given a particular hand, how likely is this player to take the action we just saw? A set will bet a wet board nearly always; a missed gutshot will barrel only sometimes; pocket jacks will check back a queen-high flop fairly often. These are likelihoods.
The posterior is the updated range after folding in the new action: your narrowed range, which becomes the prior for the next street.

The plain-language version: a hand’s share of the range goes up when it is the kind of hand that takes the action you just saw, and down when it isn’t. A player who leads big into the preflop raiser on a dry board has very few hands that want to do that, so the few that do (a slow-played monster, or a specific bluff for that player) get heavily up-weighted, while everything else nearly vanishes.

Two failure modes this framework guards against:

Common mistake

Ignoring the prior. The classic blunder is to see a big river bet, think “that bet looks really strong,” and fold top pair, without checking whether the prior range even contained enough strong hands to justify that feeling. A bet cannot represent a hand that was already filtered out two streets ago. Always update from the existing range; never reset to “well, he could have anything” the moment a scary card lands. He could not. Most of “anything” died on the flop.

The mirror error is just as costly:

Common mistake

Treating a wide action as a narrow one. Small bets, automatic continuation bets, and standard preflop calls barely narrow a range at all, yet players routinely react to a tiny flop bet as though it screamed strength. If an action is cheap and a player takes it with their entire range, that action is nearly free of information. Save your big reads for actions that cost the opponent something and that only a slice of their range would choose.

21.3 Setting the Prior: The Preflop Anchor

Your read is only as good as its starting point, so anchor carefully. Three inputs set the preflop range:

Position. The earlier the open, the tighter and stronger the range. A typical thinking player opens roughly 16–20% from middle position, 22–28% from the cutoff, and 40–50% from the button. (The 16–20% is a modern 6-max baseline; an old-school or deliberately tight reg, a regular who plays seriously and recurs at your tables, might run a few points under it. Do not default a generic “middle position” too tight, since most regs open wider than the old 12–16% guideline.) Treat these as starting estimates and bend them to the actual human.
The preflop action itself. Each action reshapes the range in a characteristic way. An open is wide and uncapped, meaning it can still contain the very strongest hands. A call of a 3-bet is condensed: it has folded the trash and usually 4-bet or folded the very top, leaving a middling band (suited broadways, medium pairs, the better suited connectors). A cold call in the middle of the action is rarely the nuts and rarely garbage. A 4-bet is small and polarized, meaning it splits into clear value hands and clear bluffs with little in between.
The player type. A 60%-of-buttons recreational player and a 38% nit, an unusually tight, risk-averse player, have wildly different priors from the same seat. Player typing, covered in the exploitative-play chapters, feeds straight into this step. When in doubt against an unknown, start from a competent-regular baseline and adjust as reads accumulate.

Key idea

Write the prior down in your head as an actual list, not a vibe. “Cutoff open from a solid reg” should resolve to a concrete set: pairs 22+, suited aces A2s+, broadways, suited connectors and gappers down to about 54s, plus the offsuit broadways and bigger offsuit aces. You cannot narrow a range you never specified.

21.4 The Three Filters on Every Street

After the prior, three filters do the narrowing on each postflop street. Run all three together; they constrain one another.

1. The line (story consistency)

The single most powerful filter is the sequence of actions itself. Real hands tell coherent stories. A player who checks the flop, then bets the turn, then overbets the river is telling a delayed-aggression story; ask which hands play that way. Often it is hands that improved on the turn, hands that were pot-controlling and then made a decision, or a specific delayed bluff. A player who bets every street with increasing size is telling a linear value or polar barrel story.

The filter question for the line is always: “Does the hand I’m imagining for them actually want to have played all the previous streets this exact way?” A surprising number of holdings you might fear get eliminated because they would have raised earlier, bet earlier, or given up by now. Story inconsistency is your friend: when a line makes no sense for value, the range tilts toward bluffs, and vice versa.

2. The bet size

Size sorts hands. The same player uses different sizes with different parts of their range, and the size they chose tells you which part.

Small bets (1/4–1/3 pot) are usually range bets, taken with almost everything on boards that favor the bettor. They barely narrow the range, so do not over-read them.
Medium bets (1/2–2/3 pot) are the standard value-and-protection size; a fairly normal, fairly wide swath of the range.
Large bets and overbets (3/4 pot to 1.5×) polarize: they tend to mean strong or bluffing, with the medium-strength hands stripped out. An overbet, especially, is a claim of “I have the top of my range or nothing.”

The caution: sizes mean different things to different players. Many recreational players invert the scheme, betting big because they have it and small as a “feeler.” Calibrate sizes to the individual; the tells in bet sizing get a chapter of their own.

3. The board texture

Texture decides which hands connect and which draws exist, and therefore which value hands and which semibluffs (bets made with a draw that can still improve) are even possible. The same line means different things on Q♠7♥2♦ rainbow (three different suits, so no flush draw: dry, few draws, almost no semibluffs possible) than on J♥T♥8♣ two-tone (two cards of one suit: wet, every other combo is a draw). On dry boards, aggression is under-bluffed by population because there is simply less to bluff with; on wet boards, the bettor’s range is full of natural semibluffs and the value claim is weaker.

Key idea

Bluffs need fuel. A player can only barrel a draw that the board actually offers. When you reach the river, count the busted draws the runout could have produced. If the texture never created many draws, the river’s big bet has few natural bluffs behind it, and a thin bluff count turns even decent made hands into folds against competent opponents.

21.5 Weights, Not Just a Yes/No List

Beginners narrow by crossing hands off a list, present or absent. Strong hand readers keep weights: a hand can be at 100% (always plays this way), 50% (does it half the time), or 10% (rarely, but possible). Weighting matters because the river decision is a ratio of value combos to bluff combos, and a hand sitting at 30% weight contributes only 30% of its combinations to that ratio.

When you update, you do two things at once: you delete hands that the action rules out entirely, and you re-weight the survivors by how likely each was to take that action. A turn check-raise might not delete top pair, but it should slash its weight, because top pair only check-raises occasionally while two pair and sets do it routinely. Full combinatorics, how many combos each holding has and how your own cards block them, gets its own chapter. Here, just remember that the final answer is a weighted range, not a checklist.

21.6 A Complete Worked Hand

Let us run the entire funnel on one hand. 100bb effective, 6-max online cash. Hero is in the big blind. The cutoff is a competent, balanced regular, a thinking TAG (tight-aggressive regular).

Preflop. The CO opens to 2.5bb. It folds to Hero in the BB, who calls with K♣Q♦. The small blind has folded. Pot is 5.5bb.

Set the prior. The CO’s opening range, for a solid reg, is roughly the top 26%: pairs 22+, suited aces A2s+, suited broadways and most suited connectors/gappers down to about 54s, plus offsuit broadways and the bigger offsuit aces. That is the starting distribution, wide and uncapped. (Hero flat-calls KQo rather than 3-betting to keep the 3-bet range tighter and to keep the CO’s many dominated bluffs in the pot. Hero’s own line is not our subject; we are reading the CO.)

Flop: Q♠7♥2♦ (rainbow). Hero flops top pair, good kicker.

The CO continuation-bets 2bb into 5.5bb (about 1/3 pot). Hero calls. Pot is 9.5bb.

Filter. This is a small range bet on a dry, queen-high board that massively favors the preflop raiser. Apply the filters: the line is a standard c-bet (no information yet), the size is tiny (barely narrows), and the texture is dry (no flush draws, and on Q♠7♥2♦ the ranks are too disconnected for any flopped gutshot; a board card cannot supply two cards to a straight run, so the most a hand like JT, T9, or 54 can have is a backdoor straight draw, three to a straight needing two more running cards). Conclusion: the range hardly narrows. Nearly the entire opening range fires this size, so we keep the prior almost intact. The discipline here is to avoid over-reading a cheap action.

Hero’s choice. Hero just calls rather than raising, keeping the CO’s air and dominated queens in the pot and avoiding bloating the pot with a one-pair hand out of position.

Turn: J♠. The board is now Q♠7♥2♦J♠. The CO bets 7bb into 9.5bb (about 3/4 pot). Hero calls. Pot is 23.5bb.

Filter: this is where real narrowing begins. The jack is an overcard to the seven and deuce but under the queen. Note that it is the J♠: the board is now two-tone (Q♠J♠), so while no flush is made yet, the turn puts live flush draws into play and hands some holdings a straight draw. A solid reg firing a large second barrel on this card is telling a strong story. Which hands take this line?

Value: strong top pair (AQ, KQ, of which Hero blocks some; QJs is now two pair), overpairs KK/AA (the CO opened and no one 3-bet, so these are always in his c-betting range; some players fast-play them with a larger flop size, but they never get filtered out), sets (QQ, 77, 22), and JJ, which just turned a set.
Semibluffs: the hands that picked up equity on the jack. KT and T9 now hold an open-ended straight draw, and every two-spade combo (A♠T♠, K♠T♠, T♠9♠, 9♠8♠, and the like) now has a genuine flush draw, four to a flush with one card to come rather than a mere backdoor. These are real semibluffs; the earlier “backdoor spade” framing no longer applies once the second spade lands.
Give-ups: missed small suited connectors and ace-high hands with no equity tend to check back or give up rather than fire 3/4 pot, so they leave the betting range.

Update: the range is now value-leaning but still contains a live block of semibluffs (the open-enders and the two-spade flush draws). Hero’s top pair is still ahead of the bluffs and the worse top pairs, behind two pair and better. A call is comfortable. Note the weights: JJ/sets are near 100% to barrel; AQ is high; pure air is now low. The turn offered a defined menu of draws (a couple of open-enders plus the spade flush draws), so the semibluff count is real but countable. File that exact menu away for the river, because every one of those hands is about to either complete or brick.

River: 5♣. Final board Q♠7♥2♦J♠5♣, a total brick. The 5♣ is not a spade, so no flush gets there, and on a scattered 2♦5♣7♥J♠Q♠ board no two-card holding can make a straight: every possible straight run would need three cards from the hand, which is impossible. (Even 64 only ever had 4-5-6-7, a four-card draw, never a made straight; 54 just makes a pair of fives.) The 5♣ does give pocket fives (55, in the CO’s 22+ opening range) a set, but 55 was an underpair on the Q♠7♥2♦J♠ turn that gave up rather than fire 3/4 pot, so it never arrived at the river as a bettor, and the betting range is unchanged. The river completes nothing and changes nothing. The CO overbets, 30bb into 23.5bb (about 1.3× pot). Hero has top pair, good kicker, and a decision.

Final filter. An overbet on a brick river is a maximally polar action: the CO is representing the very top of his range or a busted draw, with nothing medium. Build the two buckets.

Value that overbets here: the genuinely polar value is sets (QQ, 77, 22), QJ two pair, and JJ. These want to overbet for maximum value and all crush Hero’s KQ. The strong overpairs KK/AA belong in the overbetting range only at reduced frequency, if at all. The medium bluff-catchers, AQ (top pair) and most of KK/AA, are exactly the hands a balanced player checks back on a brick: worse hands almost never call an overbet and better hands always do, so betting them thin only loses value. This is the “medium-strength stripped out” principle from the bet-sizing section, applied here. Even after trimming AQ and most overpairs out of the betting range, the remaining sets, two pair, and JJ still make a healthy block of combinations.
Bluffs that overbet here: exactly the turn semibluffs that bricked: the busted open-enders KT and T9, plus the busted two-spade flush draws (A♠T♠, K♠T♠, T♠9♠, 9♠8♠, and similar) that missed when the 5♣ brought no spade. There are no rivered straights to weigh, since the board cannot make one, so the bluff bucket is precisely these busted straight and flush draws. That is a finite number of combos, but not a trivial one: the runout really did create some bluffs.

A rough combo count: the weighted river range, made concrete. You do not need full combinatorics to put numbers on the two buckets, and you are not expected to derive these counts from scratch here: combinatorics, how to enumerate each holding’s combos and net out the blockers, gets its full treatment in the combinatorics chapter. After removing the queen and king sitting in Hero’s own hand, the value that genuinely fires this size is roughly: sets ≈ 7 combos (QQ ≈ 1 with Hero holding a queen, 77 ≈ 3, 22 ≈ 3), JJ ≈ 3, and QJ two pair ≈ 6, call it about 16 “always-value” combos, plus a few KK/AA that occasionally overbet instead of checking. The bluffs are the busted draws: open-enders KT and T9 plus the missed two-spade flush draws. Perhaps a dozen draw combos exist in range, but the number that actually arrives as a river overbet is smaller, because Hero’s king blocks some of the KT and, more importantly, a real player turns only a fraction of busted draws into bluffs (the rest check and give up). Now the threshold: to balance a ~1.3× overbet against ~16 value combos, villain needs about 16 × (36 / 64) ≈ 9 bluff combos in the betting range. So the quantified question is: does this reg actually fire roughly nine busted-draw combos as an overbet here?

Now the arithmetic of the price. Hero must call 30 to win 23.5 + 30 = 53.5. Risking 30 to win 53.5, Hero needs to win about 30 / 83.5 ≈ 36% of the time to break even. For a perfectly balanced villain, an overbet of this size should be bluffed about 30 / (23.5 + 2×30) = 30/83.5 ≈ 36% of the time. The 60 is simply twice the 30bb bet, so this denominator is the same 83.5 that appeared in the pot-odds line above (53.5 + 30). That identical 83.5 is why the price Hero needs (36%) and the bluff frequency that makes him indifferent (36%) come out equal. So the whole decision reduces to one hand-reading judgment: does this player actually have enough bluffs to hit that 36%?

The honest answer for a typical competent reg is probably not. Notice that this verdict does not rest on the board being “dry”: by the river the runout had offered both straight draws and a flush draw, so the bluffs were available, and a dry-board justification would be wrong here. The fold rests instead on a population read, held with appropriate uncertainty. Real regulars under-bluff polar overbets; they give up too many busted draws and fire too few, especially with overbet sizing and especially live or at low-to-mid online stakes. So the number of bluffs that actually arrives tends to land below the ~9 combos and 36% needed for indifference, even though a perfectly balanced player would clear it. Hero folds. This is the higher-EV choice against this population.

Two honest footnotes, because hand reading is probabilistic, not certain:

Blockers cut both ways. The blockers that matter are the ones that strip combos out of the overbetting range. Hero’s Q cuts QQ (from 3 combos to 1) and QJ two pair (from 9 to 6), both of which genuinely overbet, so removing combos that crush him helps the call. Blocking AQ (with his Q) and most of KK/AA (with his K) barely helps, because those hands are checking back the brick rather than overbetting, and a blocker only improves a bluff-catch when it removes a hand from the opponent’s betting range. The K does block one of the KT bluffs, which cuts the other way. On balance the value-blocking, driven by that live QQ/QJ removal, dominates here, so Hero’s blockers nudge the spot toward a call. And contrary to a common misconception, value-side blockers help even against a perfectly balanced opponent: by stripping value combos out of the part of the range that overbets, they shift the value-to-bluff ratio of what remains and tip a borderline bluff-catch toward +EV. The combinatorics chapter formalizes this. (Note that AK is not a value blocker on this board: Q♠7♥2♦J♠5♣ leaves it busted ace-high, a bluff, so blocking it would hurt the call rather than help it.) Here the blockers make the fold closer without flipping it.
The read is player-dependent. Everything above assumed a balanced reg. Against a known over-bluffer, a player who loves the overbet button and fires busted draws relentlessly, the same physical situation flips to a clear call, because his bluff bucket is fat. The line, the size, and the texture set up the equation; the player supplies the bluff frequency that solves it.

Key idea

The river decision almost never hinges on “is my hand good?” It hinges on “is his range sufficiently bluff-heavy to beat my price?” Hand reading exists to estimate that one number (the bluff-to-value ratio), and the street-by-street funnel is how you earn a trustworthy estimate of it.

21.7 The Funnel at a Glance

Street	Action observed	Filter that mattered	Range after update
Preflop	CO opens 2.5bb	Position + player type	Wide ~26%, uncapped
Flop	1/3-pot c-bet on `Q♠7♥2♦`	Tiny size = range bet	Almost unchanged, still wide
Turn	3/4-pot barrel on `J♠`	Big size on a now two-tone board (draws live)	Value-leaning + finite semibluffs
River	~1.3× overbet on `5♣` brick	Polar size; every turn draw bricked	Value-leaning; population under-bluffs → fold

Notice the shape of the funnel: the range barely moves on the cheap flop bet, then narrows sharply once the bets get large and the action turns polar. That is typical. Cheap actions inform little; large, polar bets inform a lot. At the river, the deciding number is your read on how often this specific player turns busted draws into bluffs. Spend your attention accordingly.

21.8 Building the Habit

The method only helps if you run it every hand, including the ones you fold early. That is how the reps accumulate without risking money. A practical routine:

The instant the action reaches you, say the opponent’s prior to yourself as a concrete list.
After each of their actions, ask the one filter question (which hands actually take this, this size, on this board?) and consciously delete and re-weight.
Before any river decision, split the surviving range into value and bluff buckets, estimate the ratio, and compare it to your price.
Decide. Then, win or lose, replay the hand and check whether your narrowing was honest or whether you flinched at a scary card and abandoned your prior.

Drill

Take twenty hands from your own database or last session where you faced a river bet. For each, freeze the action before the river card and write out the opponent’s range as an explicit list of combinations with rough weights. Then reveal the river and the bet size, narrow to the final range, split it into value and bluff buckets, and compute the price you were getting. Only after committing to call or fold do you look at the showdown. Score yourself not on whether you “got it right” on that one hand (variance makes single results meaningless) but on whether your process was sound: a correct fold that happened to run into a rare bluff is still a correct fold. Do this for a week and the funnel starts running on its own.

Hand reading, done this way, stops feeling like mind-reading and starts feeling like bookkeeping: you maintain a running, weighted list, you debit and credit it with each action, and at the river you read the balance. The next chapters sharpen the individual filters: what bet sizing and timing actually reveal, how ranges get capped (lose the ability to hold the very strongest hands), and how to push the method to its limits. The funnel you just learned is the foundation they all build on.

Summary

Start every hand by writing the opponent’s prior as a concrete list from position, preflop action, and player type.
On each street, delete the hands that would not have taken this action, at this size, on this board, then re-weight the survivors.
Keep weights rather than a yes/no list: the river decision is a ratio of value combos to bluff combos.
Cheap actions barely narrow a range; large, polar bets narrow it sharply.
At the river, compare the bluff-to-value ratio against your price. The deciding input is usually your read on how often this specific player bluffs, not the board texture.