When somebody raises in front of you, most people’s heads offer two buttons: call, or fold.
The third button has been there the whole time, right next to them. But in a lot of players’ minds it has one very specific job — it is how you announce that you have a big hand. So it gets pressed a handful of times, and every time it does, everyone knows exactly what happened.
That is the real problem with a low 3-bet. Not timidity: treating a tool as a declaration. A declaration can only be honest. A tool can have a purpose.
This post is about the number itself: how the denominator is built, why the 8-12% target band looks so small, what too low and too high each mean, how to read your opponent’s side of it, and how big a sample it takes before any of it counts. Read the VPIP post first if you have not — the two ideas it sets up, range width and sample size, get used throughout here.
What a 3-Bet Is: The 3 Counts Bets, Not People
The name first. The 3 is not “the third player” and not “three times the size”. It counts the bets made before the flop:
- First bet: the big blind. It is forced onto the table, but by the rules it is the first bet of the street
- Second bet: somebody opens with a raise
- Third bet: you raise them back — that is a 3-bet
Count on from there: their re-raise is a 4-bet, yours after that is a 5-bet. So a preflop 3-bet is “a raise in response to one raise”, and that definition determines the denominator directly.
The numerator is simple: you raised. The denominator is:
Every hand where, when the action reached you, you faced exactly one raise, and that raise was not your own.
“Exactly one” is doing real work there. It is the condition RushCraft actually uses when parsing hand histories (raises_faced == 1, and you are not the last raiser), and it is how almost every tracking package defines it:
| Spot | In the denominator? | Why |
|---|---|---|
| CO opens, action reaches you on the button | Yes | The standard 3-bet spot |
| CO opens, BTN calls, action reaches you in the BB | Yes, and your raise still counts as a 3-bet | A call is not a raise. This raise is known as a squeeze; statistically it is a 3-bet |
| CO opens, BTN 3-bets, action reaches you in the BB | No | You face two raises; that would be a 4-bet |
| Everyone folds, you are first in | No | That is an open. From CO, BTN or SB it counts toward ATS, not 3-bet |
| You are in the BB, UTG opens | Yes | You have the chance to respond, and the obligation |
| Somebody limps, nobody raises, action reaches you | No | You face zero raises; raising there is an isolation raise |
Two things are worth stopping on.
First, the big blind is in the denominator, and it is the biggest single slice of it. This is unlike statistics such as cold call, which exclude the BB entirely on the grounds that its money is already dead and its position is the worst at the table. 3-bet does not exclude it, for a straightforward reason: the big blind is the seat that most often faces exactly one raise. Five players sit in front of you there, and any one of them opening puts you in the spot. A large part of what 3-bet measures is, in fact, how you defend your big blind.
Second, calling uses up an opportunity without adding to the numerator. This has the same shape as limping in the ATS post: you spend the chance, the numerator does not move, and what you bought is a pot in which you have no initiative. The difference is that calling here is far more often correct — facing an open, a call is a legitimate option, unlike a limp, which is almost always wrong. But it still pushes your 3-bet down, and when you read the number you want to know where the “low” came from.
Why It Matters: The Only Preflop Action That Takes the Initiative Back
The difference between calling and 3-betting is not how much money goes in. It is who gets to define the pot.
Call, and you accept their framework: they chose the size, they have the initiative, they usually bet first after the flop, and you are the one guessing whether to call. 3-bet, and all three of those flip.
Concretely a 3-bet does three things at once, and they can all be true together:
- It takes the pot down immediately. They fold and you collect the open raise plus both blinds without seeing a flop. This is the most common outcome of a 3-bet and its main source of income
- It builds a big pot with a strong hand. Just calling with aces volunteers to keep a hand that can win a huge pot locked inside a small one
- It clears out the players behind. Somebody opens, two or three players still to act — a 3-bet usually removes all of them. You have turned a multiway pot into a heads-up pot, and heads-up is where hand strength cashes in most reliably
There is a fourth thing, quieter but the most valuable long-term: your 3-bet frequency sets how wide everyone else can open.
The conclusion of the ATS post is that your stealing ceiling is set by the two players on your left. Turn that around: you are the person setting somebody else’s ceiling. With a player who never 3-bets sitting behind them, opening is as cheap as it ever gets, and everybody at the table can raise your blinds with K7 or Q9. A 3-bet is the natural predator of the steal, and it does not need to work every time — it only needs to exist often enough that nobody can assume you always fold.
The 8-12% Target Band, and the Six Spots It Averages Away
The target band RushCraft uses for 3-bet is:
8% to 12%.
Most people’s first reaction is “that low?” — VPIP is routinely in the twenties, ATS in the forties, and 8 to 12 looks like a rounding error next to them.
It is not small; the denominator is simply different. VPIP’s denominator is every hand dealt, ATS’s is “everyone folded and you are in late position”, and 3-bet’s is “you are facing exactly one raise” — a spot in which the large majority of your hands should not be re-raising anyway. 8-12% is not a statement about how aggressive you are; it is how often you act inside one narrow, specific spot.
But that single number has a problem: it is an average over several completely different situations. “Big blind facing an UTG open” and “button facing a CO open” are both called a 3-bet and are not the same thing at all.
The table below breaks out one of those groups: you are in the big blind, facing an open from a particular position. RushCraft’s practice mode sets its default 3-bet ranges exactly this way, split by the opener’s seat (a first-pass approximation; which hands each of the five groups actually contains is drawn out square by square in the preflop range charts):
| Who opened | Default 3-bet range | Share of all starting hands |
|---|---|---|
| UTG | QQ+, AKs | 1.7% |
| MP | QQ+, AKs | 1.7% |
| CO | JJ+, AQs+, AKo | 3.3% |
| BTN | TT+, AJs+, AQo+, KQs | 5.3% |
| SB | QQ+, AKs, AKo, A5s-A4s, KQs, QTs, JTs | 4.1% |
Three things in that table are worth a look.
First, the tightest and the widest differ by a factor of three. An UTG open means his range was already strong, so firing back with JJ runs into QQ+ a lot of the time; a button opens over 40% of hands, and that same JJ is crushing it. You are not 3-betting your own hand, you are 3-betting his range.
Second, look at why the SB row contains A5s, QTs and JTs. The first four rows are pure hand strength; only against an SB open do a few clearly-not-strong-enough hands appear. The reason is position: he opens from the small blind, you are in the big blind, and that is the one combination where you have position after the flop. Position is what lets you carry bluffs, because when you miss there are still three streets on which to give up or continue. Position does not decide how much you 3-bet; it decides whether you can 3-bet hands that are not strong enough.
Third, those five numbers average around 3%, a long way from 8-12% — and that distance is the answer. What is listed above is the value skeleton, the part where raw hand strength is reason enough to re-raise. Getting from 3% to 10% is a gap that only bluffs can fill.
One easy misreading to head off: that 3% is the big blind’s defensive value skeleton, while 8-12% is your average across every seat. The two have different denominators and should not simply be subtracted. The conclusion survives anyway, on two counts — the big blind is the largest slice of the denominator to begin with (see the first section), and in a seat where you do have position the value skeleton widens but stays in single digits: the one combination in the table above where you have position postflop (facing an SB open) is still only 4.1%. From whichever seat you count, the hands strong enough to re-raise on merit do not fill 8-12%.
Worth saying plainly: in a healthy 3-bet range, the hands that make money through raw strength are the minority. Most 3-bets earn their money in the instant the other player folds. If your 3-bet range is nothing but value hands, your number will sit somewhere between 3% and 5% — which is exactly the illness in the next section.
(Picking bluffs is not arbitrary. A5s gets chosen because the ace in your hand directly reduces the combinations of AA and AK he can hold, and because A5s that gets called can still make straights and flushes. 87o gives you neither — it blocks nothing and has no particularly good future once called.)
Too Low and Too High: One Bleeds Slowly, One Bleeds Fast
Too low: you are announcing that raising you is cheap
Falling below the target band costs you on three levels, each dearer than the last.
First, you collect fewer of the pots that would simply have been folded to you. The easiest to calculate, and the least painful.
Second, your 3-bet becomes a face-up card. Someone who 3-bets 4% of the time has a 3-bet range of roughly QQ+ and AK. Nobody has to guess: they see the 3-bet and fold hands like JJ and AQ that would otherwise be agonising decisions, because you are either aces or kings. The result is that your aces only ever win those blinds. You think you are protecting your big hands; what you are doing is guaranteeing they never get paid.
Third, and most expensively: you are forced to handle every medium hand by calling. AJ, KQ and 99 facing an open leave you with call or fold if you will not 3-bet. Calling means entering a pot where you have no initiative, and from the blinds it means speaking first on every street after the flop. None of those pots kills you in one go; they leak a little every day.
The fix for a low 3-bet is not “fight back more”, it is deciding which specific hands get promoted from call to 3-bet. Start with the strong suited aces (AJs, ATs) and hands like KQs that block something and still have a future when called, and add upward until you reach the band. This is also what RushCraft suggests when your 3-bet is too low, particularly from late position against an early open.
Too high: the 4-bets come, and your range cannot take them
The punishment for going past 12% arrives later and hits harder. Re-raise with a pile of marginal hands and, as soon as your opponents start 4-betting, you face the same question over and over: swallow the 9bb and fold, or take K9s into an enormous pot you want no part of.
The key here is the same as with stealing: the hands you 3-bet need a next move against a 4-bet. Able to 5-bet all in, able to call, or with a plan for after the flop — at least one of the three.
One thing worth knowing: 3-bet is one of the two statistics in RushCraft’s player classification that can label somebody a maniac on its own — the other is ATS. Cross the threshold, 11% at micro stakes and 13% elsewhere, and however normal the rest of your numbers are, that is the classification. The ATS lines are 50% and 60%.
VPIP and PFR get no such treatment: they have to cross together (29 with 24 at micro, 32 with 26 elsewhere). The reason is easy to see — a high VPIP without a high PFR is not a maniac, it is loose-passive, and that has its own slot in the same taxonomy. Frequency of putting money in cannot by itself separate “aggressive” from “loose”; it takes both numbers to say which.
There is sense in that design. The two that can convict alone both measure the frequency of aggressive raising, and a 3-bet is the preflop action that commits the most and is the least reversible; someone losing control there is far more dangerous than someone playing a few extra hands. Note that the 13% threshold sits one point above the top of the target band — between healthy and hazardous, 3-bet has almost no buffer at all.
The fix for too high is to pull the core back to value (AA/KK/QQ/AK), keep a small number of high-quality bluffs (A5s, small pairs and the like), and default marginal hands to calling or folding.
Reading Your Opponent: His Fold to 3-Bet, and the 70% Line
Everything so far has been about you pressing the button. But every 3-bet has somebody responding to it, and that number is Fold to 3-Bet. RushCraft’s line for it:
Target 65% or below; above 70% it is flagged as a serious leak, with a minimum sample of 30 opportunities.
70% is not arbitrary. It is roughly the point at which a pure-bluff 3-bet becomes automatically profitable, and one calculation shows why.
Say he opens to 2.5bb from the CO and you raise back from the big blind. A 3-bet out of position is usually around four times the open, so about 10bb (with position you can go smaller, around three times — out of position you need to be bigger, because you need more folds to compensate for being at a disadvantage after the flop).
- You are putting in an extra 9bb (the 1bb of your big blind was already on the table)
- The pot you are attacking is 2.5 + 1 + 0.5 = 4bb
- Break-even fold equity = 9 ÷ (9 + 4) ≈ 69%
So:
Below 69% and a pure-bluff 3-bet is no longer free money — you need some equity or a postflop plan from the hand itself to make up the difference (the full version of this “how often does it have to work” arithmetic is in how to work out pot odds).
Above 70% and every hand is a profitable 3-bet. 72o included. That stops being a decision and becomes a button.
It also explains why RushCraft treats 70% as a serious leak rather than mildly conservative play: it is not “a bit tight”, it is your opening range being harvested for free. In the other direction, the same line is a coordinate for reading opponents:
| His fold to 3-bet | What it means for you |
|---|---|
| 75% and up | 3-bet him very wide, and you can even size down |
| Around 60% | Normal. Your 3-bets need hand strength or a plan behind them |
| 45% and below | He rarely lets go. Cut the bluffs, attack with value, and be ready for 4-bets |
Your own fold to 3-bet matters just as much. The fix for a fold-to-3-bet that is too high is not “call more”, it is turning that spot into a three-way choice. RushCraft’s fold-to-3-bet drill is built exactly that way: you open, you face a 3-bet, and you choose between calling, 4-betting and folding, judged against a default 4-bet range of QQ+, AKs, AKo, A5s.
Note that A5s turns up again, and it is the only hand in that range not there on merit. 4-bets need bluffs too, for precisely the same reason 3-bets do — a 4-bet range of nothing but QQ+ lets your opponent stack off comfortably with kings every single time.
Sample Size: At 1,000 Hands the Interval Is as Wide as the Target Band
The conclusion first: 3-bet is among the slowest-converging of the common statistics, slow enough to change what you are allowed to conclude from it.
The cause is twofold. The denominator is smaller than VPIP’s (somebody has to open before you), and the base rate is only 8 to 12% — the lower the base rate, the less precision a given sample buys.
A rough estimate of how fast the chances accumulate at six-max: on average you face “exactly one raise” about once every three hands, so roughly 35 opportunities per 100 hands (faster than any single position in ATS, because it is not restricted to particular seats).
Assume a true 3-bet of 10%:
| Opportunities | Roughly how many hands | Observed number lands at | What that interval is good for |
|---|---|---|---|
| 100 | ~300 | 4% – 16% | Nothing at all |
| 350 | ~1,000 | 7% – 13% | Exactly the width of the whole target band |
| 1,000 | ~2,900 | 8% – 12% | You can start separating low, normal and high |
| 3,500 | ~10,000 | 9% – 11% | Only now is fine-tuning meaningful |
(Normal approximation of the 95% interval; the orders of magnitude are right.)
The second row is the point of the table: after 1,000 hands, the interval around your 3-bet is as wide as the target band itself. Which means “my 3-bet is 8.5%, is that low?” is a question a 1,000-hand sample cannot answer — it might genuinely be low, or you might truly be at 12% and simply not have been in the spot lately.
That is also why RushCraft’s 3-bet trend diagnosis answers “not enough data” and nothing more whenever the recent sample is under 1,000 hands. It compares your most recent 5,000 hands against your overall figure, reading direction rather than absolute value: overall inside the band with recent numbers outside it gets flagged as tilt; both outside is a persistent leak rather than a state of mind.
One more detail worth knowing. When the AI coach’s style summary decides whether you have been “recently aggressive” or “recently conservative”, the threshold differs by statistic:
| Statistic | Recent-vs-overall deviation needed to flag |
|---|---|
| ATS | 6 percentage points |
| 3-bet | 2 percentage points |
3-bet’s threshold is a third of ATS’s, not because it matters more, but because it moves around less to begin with. ATS swinging five points around a level in the forties is routine; 3-bet going from 9% to 11% is already a signal at the same sample size. (The flag also requires at least 500 genuine opportunities in the recent window, or it stays quiet.)
So the right way to use 3-bet, exactly as with ATS, is bracketing, not tuning:
- A few hundred opportunities: confirm whether you 3-bet at all. A number like 3% is a conclusion already; you do not need more data
- One or two thousand: put yourself in the low / normal / high bracket and change your behaviour per the section above
- Beyond ten thousand hands: only now does “9 or 11?” become a real question
If You Play Rush & Cash
The section on reading opponents is, once again, of no use on fast-fold tables — you fold and you move, you never play 100 hands against the same person, and the sample to read him never accumulates. That much is identical to VPIP and ATS.
But 3-bet has one property on fast-fold tables that the others do not, and it points upward: the section above says the punishment for a high 3-bet is opponents starting to 4-bet back — and that punishment requires them to have accumulated data on you. On a fast-fold table they cannot.
Play two hundred hands at a fixed table and the player on your left will notice your 3-bet frequency and start firing back. Fast-fold removes that: every hand is a new stranger, nobody is building a profile of you, and nobody remembers that you 3-bet thirty seconds ago.
This is not a licence to 3-bet wildly — a range that cannot take the heat still pays after the flop, and that cost has nothing to do with whether anybody knows you. But the “targeted retaliation” line really does drop out of the cost, so the fast-fold ceiling for 3-bet sits a little higher than the fixed-table one.
The other benefit is more practical: 3-bet opportunities accumulate quickly (about 35 per 100 hands), so the “one or two thousand opportunities” threshold above takes a few evenings. The way to do it is to export your hand history from PokerCraft first (six steps, see downloading your Natural8 hand history); import it into RushCraft and your 3-bet, your fold to 3-bet and the distance between each of them and the target band are already computed.
3-bet comes down to one sentence: how often, when facing a raise, you choose the third button.
- The denominator: only hands where you faced exactly one raise. The big blind is in it and is the largest slice; calling uses up the chance without adding to the numerator
- What the total hides: 8-12% averages six different spots. The value skeleton against an UTG open is 1.7%, against a button open 5.3% — you 3-bet his range, not your hand
- Skeleton and gap: value hands account for only about 3%; the rest of the way to 10% is where the bluffs go. Most 3-bets earn their money the moment the other player folds
- Both illnesses: too low bleeds slowly and makes your big hands transparent; too high puts you in pots you cannot handle — and 13% gets you labelled a maniac outright, with only one point of buffer
- The 70% line: past it, a pure-bluff 3-bet is free money. It is also the red line on your own number
- The sample: at 1,000 hands the interval is as wide as the target band. Bracket, do not tune
Next time somebody opens in front of you and your hand is neither strong nor trash, do not rush into choosing between call and fold. Work out which position he opened from first, then decide whether to press the third button.