The dealer turns over an Ace upcard and offers insurance: a side bet, up to half your original wager, that pays 2:1 if the dealer's hole card is a 10. It looks like a hedge. It's framed as protection. The math says it's the worst bet on the blackjack table — unless you're counting.
Insurance is the cleanest application of card counting in blackjack. It's a yes/no decision with a single threshold, and getting it right is worth more EV per occurrence than any other Illustrious 18 deviation. Here's the math and the decision rule.
What insurance actually is
Insurance is a separate bet, made independently of your main hand. You bet up to half your original wager that the dealer's hole card is a 10-value card (10, J, Q, K). If the dealer has blackjack, insurance pays 2:1 (you win 2x your insurance bet). If the dealer doesn't have blackjack, you lose the insurance bet and play your main hand normally.
It's not actually 'insurance' in the protective sense. It's a side bet on a specific event — dealer has a 10 in the hole — that happens to settle before your main hand plays out. The 'insurance' framing is casino marketing.
The math without a count
In a fresh six-deck shoe, 96 of the 312 remaining cards are 10-values (16 per deck × 6 decks). The probability that the dealer's hole card is a 10-value is 96/311 = 30.87% (we subtract one for the Ace already shown). For insurance to be a fair bet at 2:1 payout, the probability needs to be at least 1/3 (33.33%).
At 30.87% probability and 2:1 payout, the EV per dollar bet is:
- Win: 0.3087 × $2 = $0.6174
- Lose: 0.6913 × -$1 = -$0.6913
- Net EV: -$0.0739 per dollar bet — a 7.39% house edge.
Some sources quote insurance at a 7.7% house edge — that uses a slightly different deck-composition assumption (treating the shoe as infinite or not removing the Ace). The exact number varies by deck count and remaining-card adjustment, but the answer is the same: insurance is a roughly -7% bet on a fresh shoe. Catastrophically bad.
What changes with the count
Card counting tracks the ratio of high cards (10s and Aces) to low cards remaining in the shoe. When the true count is positive, the shoe is rich in 10s — which means the probability of the dealer's hole card being a 10 goes up. At some count threshold, that probability crosses 1/3 and insurance flips from -EV to +EV.
For Hi-Lo, the threshold is true count +3. At TC = +3, the shoe is rich enough in 10s that the dealer's hole card is a 10 more than 1/3 of the time. Above TC +3, every insurance bet is a +EV side bet. Below TC +3, every insurance bet is a -EV mistake.
The decision table
- True count below +3: NEVER take insurance. EV ranges from -7% to -2% depending on the exact count.
- True count at exactly +3: Borderline — close to break-even. Most charts say take it; a more conservative player skips it.
- True count above +3: ALWAYS take insurance. EV ranges from +1% per dollar bet at +3 to +10% or higher at +6 and above.
The threshold is the same regardless of deck count for Hi-Lo: +3 true count, full stop. For other counting systems, the insurance correlation differs — Omega II's threshold is +3 in its own system, Zen Count's is +3, KO's is around running count 0 above the IRC depending on deck count. If you're using a system other than Hi-Lo, look up the insurance index in your system's documentation.
Insurance and 'even money'
When you have a blackjack and the dealer shows an Ace, the dealer will offer you 'even money' — accept a 1:1 payout immediately instead of waiting to see if the dealer also has BJ (which would push your hand). Even money is mathematically identical to taking insurance for half your bet on your BJ wager.
The same rule applies: take even money only at TC ≥ +3. Below that, ride out the hand normally. Most of the time you'll win the full 3:2 payout; occasionally you'll push against dealer BJ. The expected value of riding it out is higher than the guaranteed 1:1 unless the count is strongly positive.
A note on single deck
Single-deck blackjack changes the insurance math slightly — the threshold is around TC +1.4 (insurance index 1 in some Hi-Lo charts). Single deck is rarely available with 3:2 payouts in modern casinos anyway, so the edge case is mostly academic. If you do find a 3:2 single-deck game, use a single-deck-specific Hi-Lo index card.
Why insurance is the most-cited Illustrious 18 deviation
Insurance is the highest-EV deviation in the entire I18 list. The decision is yes/no, the threshold is one number, and the EV swing per occurrence is large. Counters who get nothing else right but nail insurance every time recoup a meaningful chunk of the basic-strategy house edge in a counted shoe.
The catch: insurance is also the most-watched deviation by surveillance. Taking insurance only at high counts is the textbook 'counter tell.' Counters who care about heat sometimes skip insurance even at high counts to stay under the radar — trading the EV for the longer career.