Probe Betting in Poker: How to Attack a Missed C-Bet on the Turn

A probe bet is a bet made out of position on the turn or river by the player who is not the preflop raiser, after that raiser declined to continuation bet the flop. The check-back is the trigger. When your opponent passes up a free stab at the pot, they're telling you their range is short of strong hands, and probing takes the betting lead back from them. The short version: probe wide on low and connected turn cards, size small, and shut it down when an ace or a board-pairing card lands.

Below is the framework we use, the turn-card taxonomy from GTO Gecko's own engine, and some numbers we measured from our solved flop library on August 27, 2026. Disclosure: we build GTO Gecko.

Probe bet, donk bet, float bet: which is which

These three get mixed up constantly, and the distinction is entirely about who had the betting lead and when.

Bet type	Street	Position	What just happened
Donk bet	Flop (or any street after calling a bet)	Out of position	You lead into the preflop raiser before they get to act
Probe bet	Turn or river only	Out of position	The preflop raiser checked back the previous street
Float bet	Turn or river	In position	You called a flop bet with a weak hand, then bet when they checked

A probe can't happen on the flop, because nobody has checked back yet. That's the whole test. If you're first to act on the turn and your opponent passed on the flop, you're in a probe spot. If you're leading into someone who still has the initiative, you're donk betting, which is a different and much narrower play.

The practical difference matters more than the vocabulary. Donk bets are rare at equilibrium and easy to get wrong. Probes are common, high-frequency, and most players under-use them badly.

Why a check-back hands you the pot

Think about what a competent player does with a strong hand on the flop. They bet it. They bet top pair, they bet overpairs, they bet their sets, and they bet enough draws to stay balanced. A continuation bet is close to automatic on boards that favour the raiser.

So when they check instead, the strong part of their range mostly isn't there. Their checking range is capped: middling pairs, ace-high hands with showdown value, backdoor stuff they didn't want to bet, and total air they gave up with. Almost none of it wants to face two more bets.

The GTO Wizard analysis of a T-7-2 board puts a number on how lopsided this gets. On an 8 turn, roughly 73% of the button's range holds less than 50% equity against the big blind's range. That's not a marginal edge. Three quarters of their hands are behind, they've already shown they don't want to build a pot, and you get first crack at it.

Here's the part most players miss: you don't need a good hand to take that pot. You need a range that can credibly represent one, and the check-back is what makes your representation credible.

Which turn cards to probe: the four-axis test

GTO Gecko's engine classifies every turn card on four independent axes before it does anything else with the spot. The logic lives in our board-texture analyser, and it's a useful checklist even without the app in front of you.

Axis	Values	Probe-friendly end
Rank band	High (J, Q, K, A) / Mid (7, 8, 9, T) / Low (2 through 6)	Low and Mid. High cards hit the raiser's range, not yours
Pairing	Paired / Not paired	Not paired. A paired turn kills your fold equity
Suit	Rainbow / two of a suit / three or more of a suit	Two of a suit. Enough draws exist to make your bluffs make sense
Connectedness	Connected (all ranks inside a five-card window) / open-ended draw possible / disconnected	Connected. Straight-completing cards are the ones you can represent

Score a turn card on all four and the decision usually makes itself. A 5 on T-7-2 is Low, unpaired, keeps the board two-tone and completes straights. Probe it, and probe it wide. An ace on the same board is High, unpaired, rainbow and disconnected from everything: check and give up.

That ace read is the single most reliable rule in this post. GTO Wizard's data on the same T-7-2 board has the button holding 57% equity when an ace arrives, and they call it by far the worst card for the big blind on both equity and EV. It's the one card that turns their capped check-back range into a strong one.

Low turns are the most common single outcome, at 38.4%. That's the good news, because low cards are the ones that miss a preflop raiser's range and keep their check-back capped.

Why the turn comes low more often than you'd expect

Four ranks are High and four are Mid, but five are Low, so a low turn should show up about 40.8% of the time on an empty deck. We measured 38.4%, and the small gap is worth understanding, because it runs the opposite way to intuition.

GTO Gecko solves against a canonical set of 109 flops. That set is deliberately top-heavy: 73 of the 109 flops, or 67.0%, have a jack or better as their highest card. Only 5 of them top out at a six or lower.

High flops eat high cards. When a flop already holds a king, only three kings remain to come on the turn, and the deck that's left leans low. The effect is small, about two and a half points, but it points the right way for the out-of-position player. Probe spots come up more often than the raw rank counts suggest.

How much to bet

Probe sizing is not one number. It tracks how polarized your range is on that particular turn, and that follows straight from the four-axis test above.

Turn card	Your range is	Size
Low brick, board stays dry	Wide and merged	25-33% pot, high frequency
Straight-completing or draw-heavy	Merged, with real equity behind it	33-50% pot
Third card of a suit lands	Polarized	50-75% pot, lower frequency
High card that helps you specifically	Very polarized	75% pot and up, rare

The default is small. Against a capped range you're not trying to fold out hands that were never going to call two barrels, you're taxing the ace-high and second-pair hands that have to keep paying. A quarter-pot probe needs to work only 20% of the time to break even, and against a range that checked back the flop it works far more often than that.

Large probes have their place, and it's narrower than most people play it. GTO Wizard's T-7-2 work found the solver overbetting to 250% of the pot on an ace turn with a tiny, extremely polarized range. That's a specific structure on a specific card, not a general licence to blast. If you can't name why your range is polarized on this exact turn, bet small. Our guide to bet sizing covers how sizing and range shape lock together, and polarized versus merged ranges covers the underlying idea.

What to probe with

Build the probing range the way you'd build any betting range: value that wants to be called, bluffs that want a fold, and a clear reason for each.

• Thin value. Second pair, weak top pair, any pair that got there on the turn. These are the hands that gain most, because a capped range calls with worse far more often than it does against a bettor who has shown strength.
• Hands that improved on the turn. A gutshot that became an open-ender, a backdoor flush draw that picked up a second card. They have equity when called and fold equity when not.
• Straight and flush draws. Standard semi-bluffs, and the four-axis test tells you when the board supports enough of them to be believed.
• Blockers to their continues. If you hold a card that removes the strong part of their check-back range, your bluff gets through more often.

What doesn't belong: king-high with no draw, ace-high you plan to check down, bottom pair on a board where nothing worse calls. Those hands want a free showdown, and probing throws it away.

When to check instead

Four situations where the probe is a losing bet, and the first two are the expensive ones.

An ace or a high card lands. Preflop raising ranges are full of aces and broadway cards. That turn improves the exact range that just checked back, and your fold equity is gone.

The turn pairs the board. This happens on 18.4% of turns, and it hurts twice. Trips are now live in their range, and the hands you wanted to fold out have improved or gained showdown value. Probing into a paired turn is the most common probe leak we see.

The player behind you never folds. Fold equity is most of the value of a small probe. Against a station, probe your value hands and check your bluffs, which is the opposite of a balanced strategy and correct anyway.

You beat nothing and block nothing. No pair, no draw, no relevant blocker, on a card that didn't change anything. Check.

River probes

The river version is the same idea one street later: you check-called or checked through the turn, they passed again, and now the pot is yours to bid on. The range read is even stronger, because they've now declined twice.

Two things change. Bet sizing gets more polarized, since no equity remains and every bet is either value or a bluff. And frequency drops, because a river bluff that gets called loses the whole pot rather than one street. Pick the hands that block their calling range and give up with the rest.

Adjusting to who checked back

Equilibrium play assumes the check-back range is balanced. At a real table it usually isn't, and the direction of the error tells you how to probe.

GTO Gecko's Player Types trainer models opponents as a set of pot-relative action incentives applied across the whole game tree. The three archetypes it ships carry these adjustments:

Archetype	Incentives	What it means for your probe
Nit	bet -0.03, check +0.02, fold +0.07	Checks back more and folds far more. Probe wider, size small, expect it through
Fish	call +0.04, check -0.06	Calls too much. Probe for thin value, cut the bluffs
Maniac	bet +0.06, call +0.03, fold -0.06	Rarely checks back at all, so probe spots are rare. When one appears, they're genuinely weak

The nit line is where the money is. A player who folds 7% more often than equilibrium across every node also checks back more hands they should have bet, which caps their range harder than the solver assumes. Both errors point the same way: bet more often. Our post on drilling exploitative adjustments goes through how these profiles are built and where they're gated.

Three mistakes that cost the most

Probing every single turn. A check-back is an invitation, not a command. Observant opponents notice a 100% probe frequency within an orbit and start floating you light and raising your bricks.

Betting big to "punish" the check. The check-back already told you their range is weak. Weak ranges fold to small bets. A pot-sized probe risks three times as much to fold out the same hands.

Giving up after one probe gets called. Their turn calling range is still capped, they still don't have the initiative, and the river is where a capped range gets most uncomfortable. Plan two streets before you fire the first one.

Drilling probe spots

Reading about turn spots does very little. The reason probe betting stays a leak for so many players is that it turns up a few times a session, gets played on instinct, and never gets reviewed.

GTO Gecko is a mobile GTO poker solver and trainer for iOS, Android and browser. Its cash postflop track includes a Turn Barrels module, described in-app as covering "the critical turn street decisions", plus a Board Texture Mastery module that rotates the same spots across the flop textures above. Both run on single-raised pots, which is where probe spots live. You can also set up any flop in the solver, walk the tree through the flop check-check line, and look at the exact turn strategy for the node you're first to act in.

Postflop access comes with any active paid plan. The cash plan is $24.99 a month or $149.99 a year, all four game types together are $39.99 a month or $239.99 a year, and the Player Types profiles above sit in the Elite tier at $59.99 a month or $349.99 a year for one format. GTO Gecko is free to download on iOS and Android, and runs in a browser at play.gtogecko.com.

If you're still comparing tools, we priced the field in the best poker solver apps, and how to use a poker solver covers the study workflow that makes any of them worth the money.

Frequently asked questions

What is a probe bet in poker?

A probe bet is a bet made out of position on the turn or river by the player who is not the preflop raiser, after that raiser checked back the previous street. It takes the betting initiative back from an opponent who declined to continuation bet.

What is the difference between a probe bet and a donk bet?

A donk bet leads into the preflop raiser while they still hold the initiative, and it can happen on the flop. A probe bet only happens on the turn or river, and only after the raiser has already checked back. Probes are far more common at equilibrium than donk bets.

How often should you probe bet on the turn?

Frequently, but not always, and the turn card decides it. Low, unpaired, connected turn cards support a wide probing range. High cards, and especially aces, support almost none. Probing every turn is a bigger leak than probing too few.

What size should a probe bet be?

Small by default: 25-33% of the pot on dry, low turns, where your range is wide and merged. Increase to 50-75% when a third card of a suit lands or the turn polarizes your range. Sizes above pot belong to rare, very polarized spots.

Should you probe bet on a paired turn?

Usually not. A paired turn brings trips into your opponent's range and gives their weak hands showdown value, so your fold equity drops on both sides. The board pairs on 18.4% of turns, measured across GTO Gecko's canonical flop set.

Can you probe bet on the river?

Yes. A river probe follows a turn that also checked through. The range read is stronger because your opponent has now passed twice, but sizing should be more polarized and frequency lower, since a called river bluff costs the whole pot.