“Even money” sounds like a gift: you already have blackjack, the dealer shows an Ace, and the casino offers to lock in a 1:1 win instead of making you wait to see whether the dealer also has blackjack.
For a basic-strategy player on a normal 3:2 game, the offer is usually worse than keeping the blackjack. The reason is not superstition. Even money is the insurance proposition attached to a player blackjack, and the insurance side bet is normally overpriced when the remaining cards are not unusually rich in tens.
The important qualification is usually. A skilled card counter can reach situations where the remaining-card composition makes insurance—and therefore even money—mathematically favorable. That is why the old blanket phrase “never take even money” needs a careful exception.
What the offer actually does
Suppose you bet $100 and receive a natural blackjack. The dealer’s upcard is an Ace.
If the table offers even money, you can accept a guaranteed $100 profit immediately.
If you refuse:
- dealer has blackjack: your blackjack normally pushes, profit $0;
- dealer does not have blackjack: your blackjack receives the posted natural payout, commonly $150 profit on a 3:2 table.
So the choice is not “win versus maybe lose.” On a standard 3:2 game, it is:
take $100 for sure, or keep a wager that usually wins $150 and sometimes pushes.
That framing makes the price visible.
Even money and insurance are the same underlying bet
Insurance is usually offered when the dealer shows an Ace. The player may wager up to half the main bet that the dealer’s hole card is ten-valued.
On a $100 main bet, a full insurance wager is $50. If the dealer has blackjack, insurance commonly pays 2:1, producing $100 profit on the insurance bet.
When the player already has blackjack:
- main blackjack pushes against dealer blackjack;
- $50 insurance wins $100;
- net result: +$100.
If the dealer does not have blackjack:
- $50 insurance loses;
- main blackjack wins $150 at 3:2;
- net result: +$100.
That fixed +$100 is exactly what the even-money offer creates. Even money is therefore best understood as taking full insurance on your blackjack and settling the combination at +1 unit.
The break-even probability is one third
Insurance pays 2:1. Let (p) be the probability that the dealer’s hole card is ten-valued.
For a $1 insurance wager:
[ EV = 2p - 1(1-p) ]
which simplifies to:
[ EV = 3p - 1 ]
Break-even occurs when:
[ 3p - 1 = 0 ]
so:
[ p = \frac{1}{3} \approx 33.33% ]
If the probability of a ten-valued hole card is below one third, insurance has negative expected value. If it is above one third, insurance has positive expected value.
The player’s main-hand total does not change that threshold. What matters is the composition of the unseen cards.
A six-deck example after the visible cards are removed
A six-deck shoe contains 312 cards and 96 ten-valued cards.
Suppose the player’s blackjack is Ace-King and the dealer shows an Ace. Those three visible cards remove:
- two Aces;
- one ten-valued card.
Before considering any other exposed cards, 309 cards remain unseen and 95 of them are ten-valued.
The simplified hole-card probability is therefore:
[ p = \frac{95}{309} \approx 30.74% ]
That is below the 33.33% break-even point.
For a $1 insurance wager:
[ EV = 3\left(\frac{95}{309}\right)-1 \approx -0.0777 ]
or about -7.77% of the insurance wager in this specific card-removal example.
The exact number changes with deck count and every card already exposed. The lesson is not that insurance always has exactly a 7.77% house edge. The lesson is that ordinary composition starts below the break-even ten density.
Compare the blackjack directly
The same example can be expressed from the player-blackjack side.
If even money is taken, profit is fixed at:
[ EV_{even} = 1.00 \text{ betting unit} ]
If even money is refused, the dealer has a ten-valued hole card with probability (95/309), creating a push. With probability (214/309), the dealer does not have blackjack and the player wins 1.5 units.
[ EV_{refuse}=\frac{95}{309}(0)+\frac{214}{309}(1.5) ]
[ EV_{refuse}\approx1.0388 \text{ units} ]
In this simplified six-deck state, keeping the 3:2 blackjack is worth about 1.0388 units on average, versus 1 unit from even money.
That expected-value difference is small on one hand and meaningful over repeated decisions.
Why the offer feels better than the math
Even money removes one emotionally unpleasant outcome: receiving blackjack and then watching the dealer turn over a ten for a push.
That push feels like something was taken away. The player had a premium hand and ended with no profit.
The casino’s offer converts that discomfort into a guaranteed smaller result. Nothing deceptive is required. The trade is explicit: give up some average value in exchange for lower short-term uncertainty.
Lower variance can be a real preference, but it should not be mislabeled as higher expected value.
The card-counting exception
The insurance decision depends almost entirely on whether the unseen cards are rich enough in tens.
A card counter is specifically tracking information related to that composition. Under the Hi-Lo system, a commonly cited insurance index is around a true count of +3, although exact indices and efficiency depend on the system, rules, deck conditions, and how the count is implemented.
Wizard of Odds notes that insurance at a Hi-Lo true count of +3 or more is one of the major count-dependent strategy deviations.
This is an advantage-play exception, not a reason for a recreational player to “feel” that the shoe is ten-rich. Without a valid count, there is no composition evidence supporting the deviation.
Official rules show the equivalence clearly
New Jersey’s blackjack rules explicitly describe an even-money option when the player has blackjack and the dealer shows an Ace: the casino may offer a 1:1 payout instead of the player making the insurance wager. New Jersey’s regulation is a useful primary-source example of that procedure.
That wording is useful because it strips away the marketing feel. Even money is simply another way to settle the insurance decision.
Rules vary by jurisdiction and variant, so the exact availability of even money, insurance timing, and natural-blackjack payout must be checked on the table.
Why 6:5 changes the discussion
A 6:5 game already pays a natural blackjack less than a 3:2 game.
That does not mean a player should import the 3:2 even-money analysis blindly. Some 6:5 games may not offer the same even-money option, and the relationship between the main-hand payout and any insurance offer changes the combined settlement.
The cleaner decision is to evaluate the full posted rules before playing. The 3:2 vs 6:5 page explains why the lower natural payout is itself a major rule cost.
If the felt says 6:5, solve that table-selection problem before looking for a clever hedge on one hand.
Even money does not protect the session
A player can make the mathematically correct decision on even money and still lose heavily over the session.
One insurance decision does not control:
- bet size;
- number of hands played;
- side-bet exposure;
- doubles and splits;
- table rules;
- normal blackjack variance.
Do not use “guaranteed win” language as a reason to raise stakes, extend the session, or chase a loss. The guarantee applies only to that one blackjack settlement if the offer is accepted.
The hand you hold matters less than the unseen composition
Players often ask whether insurance is more sensible with 20, 16, or blackjack.
For the insurance wager itself, the dealer’s probability of a ten-valued hole card is the core question. Your own cards matter only because the cards you hold have been removed from the unseen pool.
That card-removal effect is real but usually small for a basic-strategy player. A counter tracks the broader composition over many exposed cards; a casual player does not gain a reliable signal just because the current hand contains no ten.
This is another reason not to turn a one-hand observation into an advantage-play claim.
A quick decision rule
For a standard 3:2 blackjack game:
- basic-strategy recreational player: decline even money;
- card counter with a validated insurance index: use the count-based decision;
- uncertain about the table payout or insurance rule: read the felt or rules before acting;
- playing 6:5 or a variant: do not assume the standard even-money comparison applies unchanged.
If you want the side-wager mechanics without the player-blackjack context, read Blackjack Insurance. If you want the counting framework, start with Card Counting Basics and True Count Conversion.
Why the exact finite-deck probability is slightly different from 4/13
A common shortcut says there are four ten-valued ranks out of thirteen ranks, so the dealer has blackjack with probability (4/13\approx30.77%).
That is a useful infinite-deck intuition, but a live blackjack hand has already removed cards.
If you hold blackjack, you have removed one Ace and one ten-valued card. The dealer’s visible Ace removes another Ace. In a six-deck game, that produces the 95-ten-cards-out-of-309-unseen-cards example above.
If your blackjack is Ace-5? That is not blackjack, so the even-money offer does not arise. If other players’ cards are exposed before the insurance decision, those cards also change the exact composition.
Card removal is therefore not a technical footnote. It is the reason the exact insurance probability is a composition-dependent quantity.
How large is the price of accepting even money?
Using the six-deck example above, refusing even money is worth about 1.0388 units while accepting is worth exactly 1 unit.
The difference is:
[ 1.0388-1.0000=0.0388 \text{ units} ]
On a $100 blackjack, that is about $3.88 of expected value for that specific state.
The player does not receive $3.88 every time. The actual result is still either the larger blackjack win or a push. The $3.88 is an average-value comparison across repeated equivalent states.
This is exactly the kind of small edge that is easy to dismiss because the guaranteed $100 feels more concrete.
Repeating the offer makes the cost easier to see
Suppose, purely for illustration, the player encountered 100 equivalent $100 blackjack/even-money decisions with the same 1.0388-unit expected value when declining.
Expected profit from taking even money every time would be:
[ 100\times100=10{,}000 ]
Expected profit from declining under the simplified state would be:
[ 100\times103.88\approx10{,}388 ]
The expected difference is about $388 across those 100 opportunities.
Real shoes do not repeat the identical card composition 100 times. The example simply converts a small per-decision edge into a scale that is easier to recognize.
Insurance can be correct even when the main hand is terrible
Because insurance is a separate wager, the quality of the main hand does not decide whether insurance is profitable.
A counter can theoretically face a situation where insurance has positive expected value even while the main hand is weak. Conversely, a natural blackjack does not make insurance favorable when the unseen-card composition is ordinary.
This separation is counterintuitive because players naturally think, “My hand is valuable, so I should protect it.” The insurance bet is not priced according to the value of the hand being protected. It is priced by the chance that the dealer has a ten-valued hole card.
Variance and expected value point in different directions here
Even money reduces variance on that settlement because the result is fixed at +1 unit.
Declining produces two outcomes:
- push for 0 profit if dealer has blackjack;
- +1.5 units if dealer does not.
The average is higher under normal composition, but the result is less certain.
A player can rationally prefer lower variance for non-mathematical reasons, but should describe the trade accurately: paying expected value to remove uncertainty.
That is different from claiming the hedge is “safer and therefore smarter.”
The table procedure is fast, which encourages automatic decisions
Even-money offers can happen quickly.
The dealer sees the player blackjack, shows an Ace, and asks for insurance or even money according to house procedure. Regular players may respond automatically because they have heard “always take the guaranteed win” for years.
A better habit is to decide the policy before the hand occurs:
- basic-strategy player on 3:2: decline;
- trained counter: follow the validated insurance index;
- unfamiliar variant: check the rule first.
Pre-deciding removes the emotional pressure of watching a premium hand sit next to the dealer’s Ace.
Do not mix up insurance payout and insurance return
If a $50 insurance wager wins at 2:1, the profit is $100 and the original $50 insurance stake is also returned.
The total amount coming back from that side wager is therefore $150, but the profit is $100.
This matters when reconstructing the even-money equivalence on a $100 main wager:
- dealer blackjack makes the $100 main blackjack push;
- $50 insurance produces $100 profit;
- net profit is $100.
The stake return is not extra profit.
Source checks for the count exception and table procedure
For the count-based exception, Wizard of Odds discusses insurance at a Hi-Lo true count of about +3, while noting that advantage figures depend on conditions. For the casino procedure, the New Jersey even-money rule explicitly permits a 1:1 even-money option on player blackjack against a dealer Ace instead of the insurance wager.
Those sources support two separate points: the procedure exists as an insurance-equivalent settlement, and composition-aware advantage play can create an exception to the basic-strategy answer.
Common even-money mistakes
“A guaranteed win cannot be wrong.” It can be lower expected value than an uncertain alternative.
“I have blackjack, so insurance must be better.” Your main hand and the insurance wager are separate propositions.
“The dealer has shown lots of small cards, so I feel a ten is due.” A count requires a systematic composition estimate, not a streak impression.
“True count +3 means every counting system uses the same exact threshold.” No. Index numbers depend on the count system and assumptions.
“The dealer recommends even money because it is good.” Dealers administer the offer; they are not your strategy service.
“A 6:5 table makes even money more attractive by definition.” The main payout and offer rules must be analyzed together; some 6:5 games do not offer the standard even-money settlement at all.
A composition test can be written without any counting system
The mathematical question behind insurance is simply:
[ \frac{T}{U} > \frac{1}{3} ? ]
where:
- (T) = unseen ten-valued cards;
- (U) = all unseen cards that could be the dealer hole card.
If (T/U) is greater than one third, a 2:1 insurance wager has positive expected value. If it is lower, the wager is negative expectation.
A counting system is a practical shortcut for estimating whether the unseen shoe has crossed that threshold without memorizing every card rank exactly. The underlying break-even condition does not depend on the brand name of the count.
The dealer Ace is already part of the information set
The insurance decision occurs after the dealer Ace is visible. That Ace has been removed from the unseen cards and should not be treated as if it were still available for the hole card.
The player’s own cards are also known removals. A blackjack containing a ten-value card slightly reduces the proportion of tens remaining, which is one reason the raw finite-deck probability can be a little lower than the simple 4/13 intuition.
Counting systems handle these visible removals automatically when the player updates the count correctly.
Even money can be a useful teaching example for expected value
The offer is unusually clean because both choices are easy to describe:
- accept: fixed +1 unit;
- decline on 3:2: either 0 or +1.5 units depending on dealer blackjack.
That makes it a good demonstration of a broader gambling principle: a guaranteed smaller payout can have lower expected value than a variable larger payout.
Expected value does not tell you what will happen on the next offer. It tells you the average value of the decision across repeated equivalent situations.
Do not turn the count exception into a reason to start counting casually
Knowing that card counters sometimes take insurance does not mean a recreational player should begin tracking only a few high cards during one shoe and call the estimate a true count.
Reliable advantage play requires:
- a defined counting system;
- accurate running count;
- deck estimation;
- true-count conversion where the system requires it;
- correct index numbers;
- enough accuracy to avoid turning small theoretical edges into errors.
If those elements are not present, the basic-strategy answer remains the more defensible decision.
Bottom line
Even money is usually a bad trade for a basic-strategy player on a 3:2 blackjack table because the guaranteed 1:1 profit is worth less on average than keeping the 3:2 blackjack and accepting the chance of a push. Insurance breaks even only when the dealer’s ten-hole-card probability exceeds one third. Card counting can create that exception; intuition cannot. Read the payout, understand the insurance rule, and judge the offer by expected value rather than by how comforting “guaranteed win” sounds.
The player decision is the focus; insurance mechanics live elsewhere
Even money is compelling because it converts uncertainty into a guaranteed-looking result on a blackjack. The decision question is whether that certainty is worth buying at the same underlying price as insurance. For a non-counter under ordinary composition, basic strategy says no because the dealer’s hole card is not a ten often enough to justify the implicit insurance wager.
For the complete side-bet offer, 2:1 payout, one-third break-even test, table timing and counting exception, use Insurance Bet.