28 Table Image & Weaponizing Reputation

By the end of this chapter you will be able to read how opponents currently perceive you and choose bluffs or value bets that exploit that perception.

Every other chapter in this book asks the same question from your point of view: what does my opponent have, and what is the best response? This chapter flips the camera around. Table image is the story your opponents tell themselves about you. Because their decisions against you run through that story, your image is a lever you can pull on every street of every hand.

The central idea is simple: your opponents are not playing against your cards, they are playing against their model of you. Two players holding the identical A♠K♥ on the identical board get called by completely different ranges depending on whether the table thinks they are a rock or a maniac. Same cards, different payouts. Once you internalize that, image stops being a vague soft skill and becomes a quantifiable edge you cultivate, protect, and spend.

Key idea

Your true frequencies determine your long-run EV in a vacuum. Your perceived frequencies determine how opponents respond to you right now. The gap between the two, between who you actually are and who they think you are, is where image-based profit lives. Your job is to open that gap and then exploit it.

28.1 How table image actually forms

Opponents build their model of you from a small, biased, recency-weighted sample. Understanding the inputs tells you which levers move the perception.

1. Showdowns, the loudest signal by far. Nothing rewrites your image like cards face-up. A single showdown is worth more to an observer than fifty folds, because it is the only fully verified data point they get. Show down 9♥6♥ after three-barreling a scary runout and you are “capable of anything” for the next hour, regardless of how nitty you’ve actually been. Show down A♠A♣ after a passive line and you become “only bets with it.” Whatever you show becomes the heading opponents file you under, and most players over-extrapolate wildly from it.

2. Visible frequencies. How often do you voluntarily enter pots? How often do you 3-bet (reraise a raise)? Do you reach showdown a lot, or are you always taking it down on the turn? Live, opponents track this fuzzily (“that guy plays a ton of hands”). Online it can be literally on a HUD (heads-up display, an on-screen stats overlay) showing your VPIP (voluntarily put money in pot), PFR (preflop raise frequency), 3-bet%, and aggression. That only applies on sites that still permit third-party tracking. On the HUD-banned sites that now lead in volume (see the online section below) there is no auto-HUD, so opponents track your frequencies as fuzzily as live players do.

3. Recent results and the heater effect. Players conflate running good with playing loose. Stack someone twice with legitimate hands and the table tags you as a wild gambler even when you’ve been a model of discipline. This misattribution is a gift: you get a loose, payable image for free, without ever putting a chip in light.

4. Demeanor, dress, and stereotype (live). Right or wrong, opponents read the old man in the windbreaker as a nit, the hoodie-and-headphones kid as a LAG (loose-aggressive) reg (a serious regular), the loud drunk as a station (a player who calls far too much). These priors are crude and often wrong, but they are the lens through which your early hands get interpreted, so they are worth knowing and worth manipulating.

5. Bet-sizing texture. A player who bombs pot constantly “feels” aggressive; a player who dribbles out small bets “feels” weak or tricky. Sizing shapes image independent of how often you’re actually bluffing.

The critical property of all five inputs: they are lagging, low-sample, and overweighted by the most recent vivid event. That lag is what you weaponize. The table’s model of you is always a little out of date and a little exaggerated, so you can be one thing while they react to another.

28.2 The two images and how each one pays

There are two archetypal images, and they pay in opposite currencies. Master the trade-off and you’ve grasped the whole chapter.

	Tight / nitty image	Loose / wild image
Opponents believe	“He only bets with it”	“He could have anything”
So they tend to	Over-fold to your aggression	Over-call / look you up
Your bluffs are	More effective (fold equity ↑)	Less effective (they snap)
Your value bets are	Under-paid (they fold)	Over-paid (they pay off)
The lever to pull	Bluff more, barrel more	Value-bet thinner and bigger, bluff less

Key idea

A tight image is a license to bluff. A loose image is a license to get paid. The single biggest image mistake in poker is using the wrong currency: bluffing into a station who thinks you’re wild, or trying to “trap for value” with a nitty image when nobody will pay you off. Spend the currency your image actually prints.

Weaponizing the tight image

Suppose you’ve shown down two premiums and folded everything else for an hour. The table has you as a rock. Now every aggressive line gains value, because your fold equity (the share of the time a bet simply makes them fold) is inflated. Concretely:

Triple-barrel bluffs on scary runouts get through. When the flush completes or the board pairs and the nit fires the river, people lay down strong one-pair hands, and occasionally even a set, because “he wouldn’t do that without it.” Pick your spot, tell a credible story, and take it down.
C-bet (continuation bet) and barrel light in single-raised pots. Your perceived range is so strong that opponents check-fold flops they’d otherwise float.
Steal relentlessly. Your blind steals and late-position opens face fewer 3-bet bluffs because nobody wants to “wake up a rock.”
The cost: your genuine value hands earn less. When you finally have aces, the action dries up. You accept thinner value, bet smaller to induce, and lean on your bluff-heavy edge instead.

Weaponizing the loose image

Now suppose you got caught barreling 9♥6♥, or you stacked someone in a coin-flip and the table thinks you’re a lunatic. People will not believe you, so stop bluffing and start cashing value:

Value-bet thinner than normal. Top pair weak kicker becomes a three-street value hand because they’ll call you down with ace-high “to keep you honest.”
Size up your value bets. A wild image lets you bet 75% and pot where a nit could only get a third-pot through. They’ve already decided you’re bluffing, so let them pay the premium.
Bet your monsters fast and big. The trap is unnecessary; the calling station is already built. Overbet your nutted hands into a player who has tagged you as a maniac.
Cut the bluffs to near zero. Your fold equity has cratered. Firing the river as a bluff into someone who thinks you’re capable of anything just donates the bet.

The symmetry is the lesson: the same hand wants the opposite line depending on your image. A K-high busted draw on the river is a profitable triple-barrel with a nit image and a mandatory check-fold with a maniac image.

28.3 Dynamic image management

Image is a resource you spend down and recharge, and it changes hands as fast as the cards turn over. The advanced skill is tracking your own image in real time and switching gears the moment it flips.

Key idea

Gear-switching is image arbitrage. Build a tight image cheaply (the folds that build it are free only because they are hands you were going to fold anyway; deliberately nitting below your correct opening and continuing frequencies to manufacture a “rock” look sacrifices real EV), then cash it in for a burst of aggression. When you get caught and the image flips to loose, switch to pure value and let them pay you off. Then fold for twenty minutes, recharge the tight image, and repeat. You continuously buy low and sell high on perception.

Advertising: paying to set up a future play

Advertising is deliberately making a -EV or marginal play now to install a false image that earns back more than it cost later. The classic is showing a bluff: you get to the river, your opponent folds, and you flash the 7♦2♣ before mucking. You “wasted” the information, but you’ve branded yourself a bluffer, and the next three times you bet big for value you get paid.

Advertising is real, but routinely overrated and overused. Be disciplined about it:

It only works against opponents paying enough attention to update. Advertising to a drunk recreational player who isn’t watching is pure cost. Against a HUD-using online reg who logs everything it works, but those players also re-adjust correctly, so the window is short.
The cost must be genuinely small relative to the future gain. Showing a cheap bluff you’d have made anyway is great. Spewing a stack to “look crazy” is almost never worth it; you’ll hit the payoff too rarely to recover the loss.
The best advertising is free advertising. Hands you were going to play anyway that happen to reach showdown and shape your image cost nothing. Let natural variance build the picture and you never pay for it.

Common mistake

“I’ll make a wild play so they’ll pay me off later.” Most deliberate advertising loses more in the moment than it ever recovers, because (a) half the table isn’t watching, (b) the ones who are watching adjust correctly and quickly, and (c) you rarely get the dream payoff hand before the image fades. Treat advertising as a rare, cheap, surgical tool rather than a default style. The disciplined player mostly lets real showdowns build the image and simply notices what it has become.

28.4 Reading how they perceive you

You cannot weaponize an image you can’t see. Build the habit of asking, every orbit: “What do they think I am right now?” Calibrate from:

Your last showdown(s). What is the most recent thing they saw you turn over? That is the heading you’re filed under.
Their adjustments. Are your steals suddenly getting 3-bet? They think you’re stealing too much (loose image). Are people check-folding to your c-bets every time? They respect you (tight image). Their behavior is a live readout of your image; read it the way you’d read a HUD.
Pool defaults. In an unknown live game you start as a mild unknown and lean nit until you do something memorable. Online, how much of a fresh start you get depends entirely on the site, covered in detail below.

A vital corollary: different opponents hold different images of you at the same time. The reg on your left who has watched you for three hours has a sharp, accurate model. The recreational player who just sat down has none and runs on stereotype. Aim each exploit at the right player: bluff the attentive nit-believer, value-own the oblivious station. There is no single “table” to play to.

28.5 Speech and demeanor (live)

Live, you carry a second channel of image: how you act. Used well, table presence and speech play shape perception and induce mistakes. A few durable principles:

Consistency is your shield. The strongest defense against giving off tells is a uniform tempo and demeanor on every action: same motion to bet, same time to act, whether you’re nutted or bluffing. A metronomic player gives observers nothing to read and quietly accumulates a solid, unreadable image.
Speech is mostly reverse-tell territory. The folk heuristic strong-means-weak, weak-means-strong (a player who taunts and acts strong is often weak, and one who sighs and acts disinterested is often strong) holds often enough to matter, which is exactly why you must assume thinking opponents know it too. Against a reg, layer accordingly; against a recreational player it frequently works at face value.
Talking to induce. A friendly, chatty “I might be making a move here” patter can goad a station into calling. The same words from a known nit are read as the reverse and induce a fold. Your speech is interpreted through your existing image; the words carry no fixed meaning of their own.
Stay inside the rules and ethics. Angle-shooting, acting or folding out of turn, and refusing to table your hand honestly at showdown cross from gamesmanship into cheating and risk floor penalties. Verbally misrepresenting your hand during play (the classic “I’ve got the nuts”) is legal table talk in most rooms only when the pot is heads-up. In a multiway pot, disclosing or discussing your holding, even falsely, influences the decisions of the other live players and can draw a one-down or a penalty, so save the lie for heads-up spots. Keep it to demeanor, tempo, and legal table talk.

Common mistake

Talking yourself into a tell. Many players are disciplined with their chips but leak everything through their face and mouth the moment a big hand hits. If you’re going to use speech, practice a single default and deviate deliberately, never reactively. An inconsistent talker is an open book.

28.6 Image online: notes, HUDs, and observed tendencies

Online strips away face, voice, and posture. What replaces them depends entirely on which site you’re playing, and as of the mid-2020s the online world has split into regimes that demand different assumptions. Two variables drive everything, and they are independent: (1) whether third-party HUDs/trackers are allowed, and (2) whether opponents see a persistent screen name they can pin a read to. Conflating those two is exactly where most players’ mental model breaks.

Regime A: tracking-permitted, named sites (parts of the ACR / PokerStars ecosystem). Here image can be quantified and persistent. Your opponents’ model of you is built from:

HUD stats: your VPIP/PFR/3-bet/c-bet/fold-to-c-bet, etc., accumulating across every hand you’ve played against that database. On these sites you cannot fold your way to a fresh start the way you can live.
Color-coded notes: “bluffs rivers,” “folds to 3-bets,” “overfolds turn.” These are explicit, persistent verdicts attached to a stable screen name.

Regime B: HUD-banned but named sites. The flagship is GGPoker, the current global volume leader. These sites ban third-party HUDs and auto-trackers, so no software reads your long-run frequencies off a display. But screen names are persistent and the client ships its own note tool, so an attentive reg still builds and keeps a hand-made read tied to your username across every session you share. The ban removes the automated-stat layer, not the human one. You deny regs the big-sample HUD, but you do not get a clean slate against a diligent note-taker who wrote “overbets the river when he gets caught” three weeks ago. Treat GGPoker regs as sharp live players who have seen you before: short on exact frequencies, long on memory of your patterns.

Regime C: truly anonymous tables. This is the Bovada/Ignition model and partypoker’s anonymous tables: IDs are hidden or rotate whenever the table breaks, so there is no persistent screen name to attach anything to. Here, and only here among the online regimes, no cross-session database forms at all, neither an automated HUD nor even a manual note, because a note would have nothing stable to point at. This is the regime where online image genuinely resets every session and is manipulated much as live image is: short samples, heavy recency bias, at most same-session reads that evaporate when the table breaks.

So before reasoning about online image, ask both questions: can they run a tracker, and can they keep a note on a stable name? Automated-database logic is strictly Regime A; human-note logic covers Regimes A and B against attentive regs; only Regime C and the recreational pool everywhere hand you the live-style fresh-slate-every-session model.

On Regimes A and B against regs, two consequences follow. First, the lever differs by what kind of database you face. On Regime A, at higher volume your perceived frequencies converge on your real ones, so the cheap “fold for an hour to look like a nit” lever weakens: the HUD shows your true VPIP regardless of the last orbit, and manipulation becomes about exploiting their notes on you rather than short-term theater. On Regime B you can still distort the frequency picture (no HUD is averaging you out), but you cannot erase a reg’s note, so the same “attack their standing read” logic applies: if he has you tagged “never bluffs the river,” you start bluffing rivers. (Only on Regime C is there no standing note, just what this opponent has seen this session.) Second, multi-tabling regs are not watching individual hands; they react to stats and the occasional note, not a single flashy bluff, so advertising barely functions against an opponent playing twelve tables.

Against the recreational online pool, on any regime, image works much as it does live: short samples, no notes, heavy recency bias, easily branded by one big showdown. Know which regime and which opponent you’re facing, and aim accordingly.

Key idea

There is no single “online.” On tracking-permitted sites against regs, image is an automated database: slow to move, and their notes on you are a fixed target you exploit by doing the opposite of what they wrote down. On HUD-banned but named sites (GGPoker, the current volume leader) there’s no auto-HUD, but persistent usernames and an in-client note tool mean attentive regs still keep a cross-session manual read, so you exploit their standing note rather than enjoying a fresh start. Only on truly anonymous tables (Bovada/Ignition, partypoker anon) and against every recreational player does online image behave like live: a tiny, recency-biased, session-only story, cheap to manipulate and fast to flip.

28.7 Worked example 1: spending a nit image

You’re 100bb deep in a live $2/5 game. For ninety minutes you’ve been a textbook rock, folding relentlessly, and your only two showdowns were K♣K♦ and A♥Q♥, both of which got there. The table, especially the thoughtful reg in the big blind, has you firmly filed as “only bets with it.” Time to spend that image.

Preflop. You open A♠5♠ (a suited wheel ace, a real hand with a blocker rather than pure air) to 5bb ($25) from the cutoff. The reg calls in the big blind. With the dead small blind, pot ≈ 10bb (10.4bb to be exact: 5 hero + 5 BB call + 0.4 dead SB; in $2/5 the $2 small blind is 0.4bb, not half a big blind).

Flop: K♠9♥4♦. The reg checks. This board crushes your perceived range: as the nit raiser, you’re supposed to have all the kings and overpairs, and the BB knows it. You c-bet 4bb (≈ 39% pot), a credible size for your image. The reg, holding something like 8♥7♥ or a weak pair, folds a lot here, but say he calls. Pot ≈ 18bb.

Turn: K♠9♥4♦6♠. Now your hand has picked up a flush draw and the board is wetter. The reg checks again. You bet 12bb (≈ 65% pot). The story you’re telling, a nit who flopped a king and is now barreling a turn that brings a flush draw, is extremely credible to an observer who has you tagged as honest. Crucially, you also have real equity: nine flush outs plus three aces give you a genuine semi-bluff, so even when called you have a hand. The reg, with one pair, is in agony against “the rock” and folds a meaningful share of the time. Say he calls again. Pot ≈ 42bb.

River: K♠9♥4♦6♠2♦. The flush bricks; you have ace-high and no showdown value. The reg checks. This is the moment the nit image pays. You’ve invested about 21bb across the three streets (5 + 4 + 12), so roughly 79bb sits behind. A bet here has exactly one job, fold equity, because you can never win at showdown. The credible value story your image promises is a polarized, nutted hand (polarized meaning you represent either the nuts or a bluff, with nothing in between): a set or two pair (here 9♦9♣, 4♥4♣, or K♦9♣), or K♥K♦. It is not “a king that got there”; top pair was never drawing to anything on this runout, and a nit doesn’t blast 1.8x pot with one pair anyway, which is part of why the small king is the wrong hand to represent.

Do the bluff math, because it’s the whole point. A pure-bluff bet of B into a pot of P breaks even when it folds out B/(B+P) of the time. Two sizes are on the table:

A large ~90%-pot bet of ~38bb (leaving ~41bb behind) needs 38 / (38 + 42) ≈ 47% folds to break even.
A full overbet shove of ~79bb (≈1.8x pot, all-in) needs 79 / (79 + 42) ≈ 65% folds.

Because you have zero showdown value, default to the cheapest size that secures the folds you need. Against the reg’s one-pair-and-worse range, a 90%-pot bet folds out essentially the same hands the shove does (a nit’s bluff-catcher lays down to either size), so the ~38bb bet buys the same fold while risking 38bb rather than 79bb. That makes it the higher-EV bluff. Reserve the full overbet shove for when it credibly folds out additional hands (thin bluff-catchers the smaller bet wouldn’t move), or when you also want to charge a value hand you’d happily get all-in. Either way, because the table has you filed under “only bets with it,” the actual fold rate sits well above even the 65% shove threshold, and that gap between your real fold equity and the break-even number is why the bluff profits. You take it down with five-high.

Now read the aftermath. If he does call and you show A♠5♠, the image has flipped. The whole table just learned the rock bluffs. Do not try that line again this hour. Switch gears to pure value, bet your next strong hands bigger, and let the new “he’s capable of bluffing” image pay them off. You’ve traded one currency for the other. The hand was never really about A♠5♠; it was about cashing in ninety minutes of folds for one big pot, then immediately re-pricing yourself for the next one.

28.8 Worked example 2: cashing a wild image

The first example spent fold equity. This one spends the opposite currency, and it’s the more counterintuitive lesson, because it tells you to do the thing every “tight is right” instinct screams against: fire three fat streets with a hand you’d normally pot-control. This is the loose-image half of the chapter’s central symmetry, worked with the same kind of numbers.

Same $2/5 game, still 100bb deep. Twenty minutes ago you got caught: you three-barreled 9♥6♥ on a busted draw, got snapped by middle pair, and had to table it. The whole table, especially the calling station two seats to your right who pays off everything, now has you firmly filed as “capable of anything.” People will not believe your bets. By the symmetry, your bluffs are worthless and your thin value is gold. Time to cash in.

Preflop. You open A♠8♠ to 5bb from the button. The station calls in the big blind. Pot ≈ 10bb (10.4bb: 5 + 5 + 0.4 dead SB).

Flop: A♥9♦4♣ rainbow. You flop top pair, weak kicker (a pair of aces, eight kicker), the textbook marginal made hand. Against a thinking reg you might check this back to control the pot. Against a station who has tagged you as wild and will call with any ace, any pair, or even king-high “to keep you honest,” you bet for thin value and protection: 7bb (≈ 67% pot). He calls. Pot ≈ 24bb.

Turn: A♥9♦4♣6♠. The turn bricks and you still hold top pair, weak kicker. Here is the fork. The textbook, image-blind instinct is pot control: TPWK (top pair, weak kicker) is the canonical check-back hand, too thin to bet for three streets and exposed to a raise. That instinct is correct against a normal opponent, and exactly wrong against a station who thinks you’re bluffing. So you do the opposite and size up: bet 16bb (≈ 65% pot), a far fatter bet than you’d ever make with this thin a hand under a normal image. He calls again with something like A♦5♦, T♥9♥, or K♣Q♣, “keeping you honest.” Pot ≈ 56bb. You’ve now invested 28bb (5 + 7 + 16); 72bb sits behind.

River: A♥9♦4♣6♠2♦. The board bricks again. You still have only top pair, weak kicker, but against this opponent that is a clean value hand, and a bluff would be suicide (he calls everything, so fold equity is zero). You overbet shove 72bb into the 56bb pot (≈ 1.3x pot). He sighs, says “I don’t believe you,” and calls with A♦5♦. Your eight outkicks his five; you drag a pot more than triple what a cautious line ever makes.

Now do the value math, the mirror image of the bluff math. Model the station as a player who calls the river with essentially his whole range (that’s what makes him a station). Let e be your equity against that calling range, roughly the fraction of his calls that are worse than your top pair. Comparing a river bet of size B into pot P against simply checking it back:

\[\text{EV(bet)} - \text{EV(check)} = B \times (2e - 1).\]

The break-even is e = 50%: as long as more than half the hands the station calls with are worse than your TPWK, betting beats checking, and every extra chip of B multiplies the edge. That single fact is the whole mathematical case for “value-bet thinner and bigger” against a station: when e > 50%, size is your friend, so you overbet.

Plug in numbers. A station calling down with hands like A♦5♦, T♥9♥, K♣Q♣, and the rest of his weaker aces, second pairs, and ace-high junk is worse than your top-pair-aces maybe two-thirds of the time, so e ≈ 0.67 and (2e − 1) ≈ 0.34. The 72bb overbet therefore nets about 72 × 0.34 ≈ +24bb over checking the river, on that street alone. A timid, nit-sized ~20bb river bet (the size a normal image is stuck with) would net only 20 × 0.34 ≈ +7bb. Same hand, same opponent, roughly three times the profit purely from sizing up, because the per-chip edge (2e − 1) is fixed, so you want to apply it to as many chips as the station will still call.

And that compounding is why you barreled the turn rather than pot-controlling. Betting the turn for value grew the pot from ~24bb to ~56bb, so the river overbet’s 0.34 edge is leveraged over a 56bb base instead of a 24bb one. The fully passive line (bet flop, check turn, check river) forgoes both streets of this edge; against a station who would have called, that’s on the order of +25-30bb of EV left on the table. Against this opponent, checking down your thin made hands is the real leak.

Two honest caveats, both of which only sharpen the lesson. First, a real station folds his very worst hands sometimes; when he does, you lose a sliver of thin value, but the hands that do call are then even more concentrated in “worse than you,” so e among callers rises and the bet stays good. Second, size to his call threshold rather than to infinity: a 1.3x overbet a station snaps off beats a 3x overbet that finally makes even him fold. The goal is to maximize B that still gets called, the value-side analogue of the nit example’s “bet the size that does the job.”

Now read the aftermath. This time you want him to call, and you showed the goods: top pair, three fat streets, dragged. Note what that does to your image: nothing useful. He already thought you were wild, so getting paid off light just confirms it, and the wild image stays live for the next strong hand. So keep spending the same currency: keep value-betting thin and big, keep your bluffs near zero, until a showdown finally rebrands you as solid. Only then do you flip back to the nit playbook of Example 1. The two examples are the same engine run in opposite gears: one cashes folds, the other cashes calls, and your only job is to read which currency the table is currently handing you.

Drill

For your next session, track your image explicitly. Every time you reach a showdown (yours, or one you’re tagged by), jot one word for how the table now sees you: nit, normal, wild. Then, before each river decision, ask in your head: “Given my current tag, am I being paid to bluff or to value-bet?” Take the line that spends the currency your image is actually handing you. After the session, review every river bluff you fired into a wild image and every thin value bet you skipped into a nit image. Those two error buckets are where image leaks hide.

28.9 Summary

Opponents play against their model of you, not your cards. The gap between your real and perceived frequencies is your edge.
Image forms from showdowns (loudest), visible frequencies, recent results, demeanor, and sizing, all of which are lagging, low-sample, and recency-biased.
A tight image is a license to bluff; a loose image is a license to get paid. Spend the right currency; the same hand wants opposite lines under opposite images.
Gear-switch dynamically: build a tight image cheaply, cash it in for aggression, and the moment it flips to loose, switch to pure value.
Advertising is real but overrated: keep it cheap, surgical, and aimed only at opponents who are watching and able to update.
Read your image off their adjustments; remember different opponents hold different images of you at once.
Live, image runs through demeanor and speech: prize consistency, treat speech as reverse-tell territory, and stay within the rules.
Online, the model depends on the site: on tracking-permitted sites against regs, image is an automated, persistent database, so exploit their notes on you; on HUD-banned but named sites (GGPoker, the volume leader) regs lose the auto-HUD but keep persistent usernames and manual notes, so a cross-session manual read still forms and you exploit their standing note; only on truly anonymous tables (Bovada/Ignition, partypoker anon) and against the recreational pool everywhere does no cross-session database form and image behave much like live.