8 MDF, Alpha & Optimal Bluffing Ratios
Every bet in No-Limit Hold’em asks the same question of your opponent: call, raise, or fold? The branch of poker theory in this chapter answers a deeper question hiding underneath that one — how often must they continue so that you cannot simply bet any two cards and print money, and how often must your own bets be bluffs so that they cannot simply call you down with anything?
These two ideas — Minimum Defense Frequency (MDF) and the optimal value-to-bluff ratio — are two sides of one coin. Both fall directly out of pot odds and the size of the bet. Master them and you will understand why solvers bet the sizes they bet, why certain rivers get jammed and others get checked, and how to spot the exact moments your opponents drift away from balance so you can punish them. This chapter derives every number from scratch so you are never memorizing a table you don’t understand.
8.1 The bluff that always works (and why defense exists)
Imagine an opponent who folds 70% of the time to a river bet. You hold the absolute worst hand possible — say 7 2 on a board that missed you entirely. Should you bluff?
Suppose the pot is 100 and you bet 100 (a pot-sized bet). You risk 100 to win the 100 already in the middle.
- 70% of the time they fold and you win 100.
- 30% of the time they call and you lose your 100 bet.
Expected value of the bluff:
\[ EV = (0.70 \times 100) + (0.30 \times -100) = 70 - 30 = +20 \]
A bet with the worst hand in the deck prints +20. That is a catastrophe for the defender. If betting trash is profitable, your opponent can bet every single hand they ever have, and you are being robbed.
The defender’s job, then, is to call (or raise) often enough that this pure bluff breaks exactly even. The frequency at which a pure bluff’s EV equals zero is the Minimum Defense Frequency.
MDF is the minimum fraction of your range you must continue with (call or raise) so that an opponent betting a stone-cold bluff makes exactly zero profit. Defend at least this often and “bet any two cards” stops working.
8.2 Deriving MDF and alpha
Let the pot be \(P\) and the bet be \(B\). The bettor risks \(B\) to win \(P\). If the defender folds with probability \(f\) and continues with probability \((1-f)\), the EV of a pure bluff is:
\[ EV_{bluff} = f \times P \;+\; (1-f)\times(-B) \]
Set that equal to zero to find the fold frequency that makes bluffing break even:
\[ f \times P = (1-f)\times B \] \[ fP = B - fB \] \[ f(P + B) = B \] \[ f = \frac{B}{P + B} \]
That quantity — the fold frequency that makes a bluff indifferent — is called alpha (α):
\[ \boxed{\alpha = \frac{B}{P+B}} \]
The defender must therefore continue at least \(1 - \alpha\) of the time. That continuation frequency is the Minimum Defense Frequency:
\[ \boxed{MDF = 1 - \alpha = \frac{P}{P+B}} \]
Notice the elegant symmetry. Alpha is the fraction you may fold; MDF is the fraction you must defend, and they sum to one. Alpha is also exactly the pot odds the bettor is laying themselves on a bluff — which is why the same fraction governs both sides of the bet.
Let’s confirm with the pot-sized example. \(P = 100\), \(B = 100\):
\[ \alpha = \frac{100}{200} = 0.50, \qquad MDF = \frac{100}{200} = 0.50 \]
So against a pot-sized bet you must defend 50% of your range. In our opening scenario the opponent folded 70% — far more than the 50% ceiling — so bluffing trash was hugely profitable. Tighten that defender up to folding only 50% and the bluff EV becomes:
\[ (0.50 \times 100) + (0.50 \times -100) = 0 \]
Exactly break-even. The leak is sealed.
8.3 The MDF reference table
Because MDF depends only on bet-size-as-a-fraction-of-pot, you can tabulate it once and use it forever. Bet size is expressed as a fraction of the pot.
| Bet size (× pot) | Alpha (may fold) | MDF (must defend) |
|---|---|---|
| 0.25× | 20% | 80% |
| 0.33× | 25% | 75% |
| 0.50× | 33% | 67% |
| 0.66× | 40% | 60% |
| 0.75× | 43% | 57% |
| 1.00× (pot) | 50% | 50% |
| 1.50× | 60% | 40% |
| 2.00× | 67% | 33% |
Read it like this: against a half-pot bet you must continue with 67% of your range; against an overbet of twice the pot you only need to keep 33%. The bigger the bet, the more you are allowed to fold — but the more each individual bluff costs the bettor when called.
Small bets demand wide defense; large bets permit tight defense. A 25%-pot stab must be called down with 80% of your range, which is why over-folding to tiny bets is one of the most common and most expensive leaks in poker.
8.4 The other side of the coin: optimal value-to-bluff ratios
MDF tells the defender how often to continue. The mirror question faces the bettor: when you build a polarized betting range of value hands and bluffs, what proportion should be bluffs so that a “bluff-catcher” — a hand that beats your bluffs but loses to your value — is made exactly indifferent between calling and folding?
If the caller is indifferent, they cannot exploit you no matter what they do. That is the goal of a balanced betting range.
Here we reason from the caller’s pot odds. When you bet \(B\) into a pot of \(P\), the caller must call \(B\) to win the pot plus your bet, \(P + B\). Their required equity to call is:
\[ \text{equity needed} = \frac{B}{P + B + B} = \frac{B}{P + 2B} \]
The caller is indifferent when their bluff-catcher wins exactly that often — i.e., when the bettor’s range contains that fraction of bluffs. So the optimal bluff fraction of the betting range is:
\[ \boxed{\text{bluff share} = \frac{B}{P + 2B}} \]
Deriving the pot-sized river ratio
Take a pot-sized river bet, \(P = 100\), \(B = 100\):
\[ \text{bluff share} = \frac{100}{100 + 200} = \frac{100}{300} = \frac{1}{3} \]
One third of the betting range should be bluffs, two thirds value — a 2:1 value-to-bluff ratio. For every two value bets you fire on a pot-sized river, you want one bluff.
We can sanity-check from the caller’s seat. Facing a pot bet they’re getting 2-to-1 (call 100 to win 200), so they need 33% equity. If your range is 2 value : 1 bluff, their bluff-catcher beats exactly the 33% that are bluffs. Calling wins 33% of the time for the breakeven they were quoted. Indifferent — perfectly balanced.
Value-to-bluff by river bet size
The same formula, applied across sizings, produces the canonical river table. (These ratios assume the river is the final bet, so the bluff has no further fold equity to collect.)
| River bet (× pot) | Bluff share of range | Value : Bluff |
|---|---|---|
| 0.33× | 20% | 4 : 1 |
| 0.50× | 25% | 3 : 1 |
| 0.66× | 28.5% | ~2.5 : 1 |
| 0.75× | 27% → 30% | ~2.3 : 1 |
| 1.00× (pot) | 33% | 2 : 1 |
| 1.50× | 37.5% | ~1.7 : 1 |
| 2.00× | 40% | 1.5 : 1 |
| 3.00× | 43% | ~1.3 : 1 |
The pattern: the bigger you bet, the more bluffs you are entitled to. An overbet of 2× pot can be 40% bluffs and still leave the caller indifferent. This is the engine behind the polarized overbet — large sizing both charges your value maximum and licenses a heavy bluffing wing.
Players memorize “2:1 on the river” and apply it to every bet on every street. The 2:1 figure is specific to a pot-sized river bet. Smaller bets want fewer bluffs (3:1, 4:1); bigger bets want more (1.5:1). And on earlier streets the math changes entirely — see below.
8.5 Why earlier streets need more bluffs
A river bet is the last bet; a turn or flop bet is not. When you bet the turn, some of your “bluffs” are actually semi-bluffs with equity, and more importantly you retain the right to fire again on the river. Both effects mean your betting range can hold more bluffs on earlier streets than the pure river formula suggests.
The classic shortcut for multi-street balance: on the flop you can run roughly 2 bluffs for every 1 value hand, on the turn roughly 1:1, and by the river roughly 1 bluff for every 2 value hands (the 2:1 value:bluff we derived). The intuition is that a fraction of bluffs get to give up on each street, so to arrive at the river with the correct 2:1 value-heavy mix, you must begin earlier with a far bluff-heavier range. Think of it as a funnel: bluffs are filtered out street by street as the unproductive ones check back or fold, while value hands keep betting.
This is why a flop continuation-betting range can look almost reckless — loaded with backdoor draws and overcards — yet be perfectly sound. Those hands are not all meant to reach showdown betting; they are the raw material from which a balanced river emerges.
Balance is a destination, not a single street. Bluff-heavy early, value-heavy late. A correct river range with ~33% bluffs is the residue of a flop range that may have been ~67% “bluffs,” most of which fold out along the way.
8.6 Polarization, sizing, and how they lock together
Three quantities move together every time you choose a bet size:
- Required defense (MDF) — falls as you bet bigger (\(MDF = P/(P+B)\)).
- Permitted bluffs — rises as you bet bigger (\(B/(P+2B)\)).
- Polarization — how value-or-bluff your range is.
Big sizings are only correct with a polarized range — strong value hands and nothing-hands — because a polarized range wants to charge the maximum and can afford the extra bluffs. Medium-strength hands (your bluff-catchers and thin value) prefer small sizings or checks, because they want to keep weaker hands in and avoid being raised off their equity. This is the deep reason solvers split ranges by size: the size is chosen to match the shape of the range betting it.
A quick mental model:
- Want to bet big? Your range had better be polar (nuts + air). You’ll defend less when checked to, but you owe a heavier bluff count.
- Want to bet small? Use a condensed/merged range (lots of medium-strong hands). You’ll bet frequently, owe fewer bluffs, and the opponent must defend very wide.
8.7 A fully worked river example
Stacks are deep; we’re heads-up on the river. The board reads:
K♠ 9♦ 4♣ 7♥ 2♠ — a dry, draw-free runout.
The pot is 60bb. You are in position with a betting range built across the hand. You decide to bet 60bb — pot-sized. How should your range look, and how should your opponent respond?
Your side — value-to-bluff. A pot-sized river bet wants 2 value : 1 bluff (33% bluffs). Suppose your value region here is roughly: sets (KK, 99, 44, 77, 22 where reachable), two pair, and strong top pairs like AK, KQ — call that 12 combos of value. To stay balanced you may add 6 combos of bluff (12 ÷ 2 = 6). Natural candidates are missed hands with the least showdown value — busted gutshots and the like, e.g. some Q-J, J-T, T-8 type holdings that have no chance of winning at showdown. You deliberately bluff your worst hands and check back hands with modest showdown value (like a weak pair) that can win unimproved.
Total betting range: 12 value + 6 bluff = 18 combos, of which 33% are bluffs. Correct.
Their side — MDF. Facing a pot-sized bet, your opponent’s MDF is:
\[ MDF = \frac{60}{60 + 60} = 50\% \]
They must continue with half of the range they arrived at the river with. They will call with their best bluff-catchers — hands that beat your 6 bluff combos but lose to your 12 value combos, such as a weaker king (KT, KJ) or 9x. They should fold their air, and they should defend exactly enough that you cannot profitably bluff your 7 2 equivalent.
Putting it together. If they defend the required 50%, your 6 bluffs break even (they win the pot half the time, lose half the time — net zero), and your 12 value bets get paid by the half of their range that calls. You profit from value; your bluffs are free insurance that they can’t simply fold everything. The system is in equilibrium: neither of you can deviate to gain.
Now the exploit. Suppose this particular opponent is a calling station who defends 75% against this bet rather than 50%. MDF says you only needed them to defend 50%, so they are over-defending by 25%. Against that player you slash your bluffs — drop the 6 bluff combos toward zero — and bet your value even thinner, adding hands like K-T for value because they’ll pay off worse kings. Conversely, against a nit who folds 70%, you fire every bluff you can find, because each one prints exactly the +20 we computed at the very start of this chapter.
MDF and optimal ratios describe the GTO baseline — the strategy that cannot be exploited. They are not the maximally profitable strategy against a specific, flawed opponent. Use them as your default and your reference point, then deviate deliberately the moment you have a read. Defending “by the chart” against a player who never bluffs just donates chips.
8.8 Practical caveats and limits
A few honesty checks before you weaponize these numbers:
- MDF assumes the bettor can bluff with any two cards. When your opponent’s range is capped or uncapped in a way that limits their bluffs (for example, a passive player who would have raised earlier with strong hands, or a board where few bluffs are even possible), strict MDF over-defends. You may fold more than MDF against opponents who structurally can’t have enough bluffs.
- Card removal and blockers matter. The clean combo counts above ignore the fact that your own cards block some of your opponent’s value or bluffs. On real boards, choosing which bluffs to fire is governed by blockers — prefer bluffs that block their calling/value range and unblock their folds. (Blocker theory gets its own treatment elsewhere in this book.)
- Multiway pots break the heads-up math. MDF as derived is a heads-up concept. With multiple defenders the collective defense, not each individual’s, must reach the threshold, so each player defends less.
- Stack depth and future betting. The river ratios assume no future streets. On the flop and turn, implied odds, semi-bluff equity, and the threat of further barrels all shift the correct frequencies, as discussed above.
For each bet size below, state (a) alpha, (b) MDF, and (c) the optimal value:bluff ratio for that size as a river bet. Do them in your head, then check against the tables.
- Half pot (0.5×)
- Three-quarter pot (0.75×)
- Pot (1.0×)
- Double pot (2.0×)
Answers: (1) α 33%, MDF 67%, ~3:1. (2) α 43%, MDF 57%, ~2.3:1. (3) α 50%, MDF 50%, 2:1. (4) α 67%, MDF 33%, 1.5:1.
Pot is 80bb. Villain bets 20bb (a 25%-pot stab) on a blank river. (a) What is your MDF? (b) You hold a middling bluff-catcher that beats only his bluffs. He is a typical tight-aggressive regular who rarely bluffs this tiny on a blank. Do you defend by MDF or fold? Explain in one sentence.
Answer: (a) MDF = 80/(80+20) = 80% — you must continue with four-fifths of your range against such a small bet. (b) Against a player whose tiny blank-river bets are almost all thin value, his bluffs fall short of the 20% required, so you may correctly fold more than MDF allows — MDF is the floor versus a balanced bettor, not an obligation versus an under-bluffer.
8.9 Summary
- Alpha \(= B/(P+B)\) is the fraction a defender may fold; MDF \(= P/(P+B)\) is the fraction they must defend. They sum to 1.
- Small bets force wide defense (80% vs a quarter-pot bet); big bets permit tight defense (33% vs a 2× overbet).
- The optimal bluff share of a betting range is \(B/(P+2B)\). A pot-sized river bet wants 2 value : 1 bluff; smaller bets want more value (3:1, 4:1), bigger bets want more bluffs (1.5:1).
- Be bluff-heavy early, value-heavy late — balance is the river destination, reached by funneling a bluff-rich flop range down street by street.
- Bet big with polar ranges, small with merged ones; sizing and range shape are chosen together.
- These are the unexploitable defaults. Once you have a read — a station who over-defends, a nit who over-folds — abandon the chart and exploit, then return to the baseline when the read runs out.