Never take insurance — the actual math

The dealer slides out an ace and asks the table: "Insurance?" The word is doing a lot of work. Insurance sounds prudent. Responsible, even — you're protecting your hand against the blackjack that might be sitting under that ace. Half your bet now to avoid heartbreak in three seconds.

Here's the thing your gut needs to hear: insurance has nothing to do with your hand. It's a completely separate side bet on one question only — is the dealer's hole card a ten? It pays 2 to 1 if yes, and takes your money if no. Your 20, your 16, your pair of aces: irrelevant. Once you see it as a standalone bet on a hidden card, you can settle it by counting ranks.

The whole argument fits on one hand

A deck has thirteen ranks. Exactly four of them — 10, Jack, Queen, King — turn the dealer's ace into blackjack. So in a fresh multi-deck shoe, the hole card is a ten-value roughly 4 times in 13, about 30.8% of the time.

A bet that pays 2 to 1 breaks even when it wins 1 time in 3 — 33.3%. Four out of thirteen is less than one out of three. That's the entire proof. The bet simply doesn't win often enough to justify its price.

Run the numbers per dollar of insurance: you win $2 about 4/13 of the time and lose $1 about 9/13 of the time. That nets out to roughly −7.7 cents on every dollar, every time you take it.

Played with basic strategy, the game itself costs you around half a cent per dollar. Insurance costs about 7.7 cents. It's roughly fifteen times worse than the game you sat down to play.

"But I have a 20!"

This is where the gut digs in: my hand is too good to lose to a blackjack. Feel the logic of that for a second — then notice it's backwards. Your 20 is made of two ten-value cards. Those are two of the exact cards the dealer needs in the hole for your insurance bet to win. You're holding the evidence against the bet you're about to make. If anything, a 20 makes insurance slightly worse than average, not better.

The strength of your hand never enters the insurance math. There is no hand you can hold that turns a 30.8% shot into a 33.3% one.

"Even money" is the same bet wearing a tuxedo

The sneakiest version comes when you have blackjack and the dealer shows an ace. The dealer offers "even money" — take a guaranteed 1× payout right now instead of risking a push. Almost everyone takes it. It feels like free money.

It's insurance in disguise, and the math is the same. Decline, and one of two things happens: the dealer has blackjack (about 4/13) and you push, or they don't (about 9/13) and your blackjack pays 3:2. That works out to about 1.04 per unit bet on average — against the flat 1.00 you locked in by taking even money.

Yes, declining means you'll sometimes push the prettiest hand in the game and it will sting. The certainty of even money is real — it's just not free. You're paying about four cents per dollar for it, forever, on the best hand you can be dealt.

The honest footnote

Is insurance always bad? Almost. The bet becomes profitable only when the remaining shoe is unusually rich in tens — which is exactly the information card counters track, and it's why the counting literature treats insurance as one of the most valuable deviations in the game. But that's a statement about a tracked shoe, not a fresh one. If you're not counting, you don't have that information, and the fresh-shoe math above is the math you're playing against. Like 16 vs 10, the play that feels protective is just the slower way to pay.

The takeaway

• Insurance is a side bet on the hole card — your hand has nothing to do with it.
• It needs to win 1-in-3 to break even; it wins about 4-in-13. That gap is ≈7.7 cents per dollar, every time.
• "Even money" on your blackjack is the identical bet with better branding — declining is worth about 4% more over the long run.
• Casinos don't market losing bets as "insurance" by accident. When a bet is named after a feeling, check the math twice.