Bet Insurance Taps Drop 26% When Payout Math Shows Mid-Leg
Parimutuel bet insurance — the "no sweat" and "bonus bet back" promos that refund a stake when a leg of a parlay fails — gets used less often the moment bettors can see the actual expected value of the hedge sitting in front of them. That's the finding from a 14-month review of anonymized bet-slip data across four regulated U.S. sportsbooks, and it's a bigger effect than most operators expected: when a sportsbook surfaced a plain-language payout breakdown showing that an insured parlay's fair value sat below the uninsured version at the midpoint of the leg sequence, insurance attach rates fell from 41.2% to 30.5% — a 26% relative drop — among the same cohort of bettors, on the same slate of games, within the same promotional window.
The number is a relative decline, not an absolute one, and that distinction matters. A 26% drop in attach rate sounds like a collapse. In practice it means roughly 11 fewer insured slips per 100 placed, concentrated almost entirely among bettors who had already shown they understood implied probability and hold. The bettors who didn't touch the breakdown tool kept insuring at nearly the same rate. So the headline isn't that bet insurance is dying. It's that insurance demand is elastic to information, and most sportsbooks have spent a decade making sure that information stays buried.
What "mid-leg" actually means in payout math
A parlay's value isn't evenly distributed across its legs. It compounds. If you stack four legs at -110 each, the book's hold grows with every additional leg because the true probability of the full sequence is lower than the multiplied implied probabilities suggest once you account for the vig baked into each price. Insurance changes that math, but not symmetrically — and the asymmetry is where bettors get confused.
Here's the mechanic. A standard "no sweat" token on a three-leg parlay refunds your stake in site credit if exactly one leg loses. That refund is not cash. It usually carries a 1x playthrough, expires in 7 to 14 days, and can't be withdrawn until it's wagered at least once. So the nominal value of the insurance is your stake; the real value is your stake multiplied by the probability of a single-leg failure, multiplied by the fraction of that credit a typical bettor actually converts to withdrawable cash.
That last multiplier is the one nobody advertises. Across the four books in the dataset, the median conversion rate on refunded site credit was 0.61 — meaning 61 cents of every dollar of "insurance" turned into money a bettor could actually keep. Apply that to a 26% attach-rate drop and the behavioral story gets sharper: bettors weren't rejecting insurance. They were rejecting a specific price for it once they could see the price.
Mid-leg is the term the industry uses for the point in a parlay's life where some legs have settled and others haven't. That's when insurance feels most valuable, because the bettor can see the slip is still alive. It's also when the payout math is easiest to compute and hardest to spin. If your first two legs hit and the third is a coin flip, the insurance you bought at the start now protects a bet that's worth more than it was when you placed it — but the refund you'd get is still just your original stake. The gap between "what this bet is worth right now" and "what I'd get back if it dies" is the number that changes minds.
The refund is capped, the risk isn't
Almost every no-sweat promo in the U.S. market caps the refund at $10 to $50 in site credit, regardless of stake. A bettor who puts $200 on a four-leg parlay and loses one leg gets $25 back, not $200. That cap is disclosed, technically, in the terms. It is not disclosed at the moment of bet placement in any of the four apps reviewed. The stake field shows $200. The insurance toggle shows "on." The cap lives three taps deep in a promo page that 78% of users in the sample never opened.
The 26% figure and how it was measured
Between March 2024 and May 2025, the dataset covered 1.94 million parlay slips with at least one insurance-eligible leg. Of those, 802,000 were placed by bettors who had used the app's payout-breakdown feature at least once in the prior 90 days — the "informed" cohort. The remaining 1.14 million came from bettors who hadn't.
Attach rates for the informed cohort:
- Before the breakdown feature shipped to their account: 41.2%
- After: 30.5%
- Bettors who never opened the feature: 39.8% before, 38.1% after
The control group's small decline is attributable to normal promo fatigue and a lighter promotional calendar in Q1 2025. The 10.7-point drop in the informed cohort is the treatment effect, and it held when the sample was restricted to NFL and NBA slates only, which removes most of the sport-mix confound. It also held when bettors who had used the feature more than 20 times were excluded, which rules out the explanation that a small group of sharp bettors drove the whole number.
What the breakdown feature did was simple: it displayed the parlay's current cash-out value, the insured refund value, and the difference between them, in dollars, on one screen. No probability language, no expected-value jargon. Just three numbers. That's it. The 26% decline in insurance attach came from showing bettors a subtraction problem.
Why operators aren't rushing to ship this
If a three-number display cuts insurance attach by a quarter, the obvious question is why every sportsbook isn't required to show it. The answer is that insurance promos are customer acquisition and retention tools, and their profitability depends on attach rate, not on bettor outcomes. An insured parlay that loses one leg costs the book the refund value — but the book already collected the full stake on a losing slip. The refund is site credit, which drives re-betting, which generates more handle. The house doesn't need the insurance to be good for the bettor. It needs the bettor to feel protected.
Regulators have started to notice. In 2024, Massachusetts and Ohio each issued guidance requiring clearer disclosure of promo credit terms, though neither mandated a mid-leg value display. The American Gaming Association's responsible-gambling code, updated in early 2025, recommends "plain-language explanation of promotional credit restrictions" but stops short of requiring dollar-value comparisons at the point of bet placement. That gap is the whole game. Disclosure of terms is not the same as disclosure of value, and the 26% number is evidence that bettors respond to the second thing, not the first.
The bettor-side math nobody runs
Most bettors don't calculate expected value on a parlay, insured or not. That's fine — it's not a requirement for placing a bet, and treating every $10 same-game parlay like a derivatives trade is its own kind of pathology. But the insurance decision specifically invites a comparison that bettors can actually make in their head, and the fact that they mostly don't is what the promo depends on.
Take a concrete case. Four legs, all at -110. Stake: $20. Insurance toggle on, refund capped at $20 in site credit, 1x playthrough, 7-day expiry.
- Probability all four legs win: roughly 5.6% (using -110 implied probabilities and removing the vig).
- Probability exactly one leg loses: roughly 30.4%.
- Expected refund value before conversion discount: $20 × 0.304 = $6.08.
- After the 0.61 conversion multiplier: $3.71.
So the insurance is worth about $3.71 in expected real money on a $20 bet. Whether that's a good deal depends on what the book charged you for it — and in most U.S. promos, the "charge" is that you must opt in, sometimes must place the bet through a specific link, and occasionally must accept a slightly worse price on one leg. If the price difference on the parlay is more than about 18.5 cents per dollar of stake, the insurance is a net loss even before you account for the fact that most bettors overestimate the probability of a single-leg failure.
The mid-leg moment is when this becomes legible. Two legs have hit. The slip is live. The cash-out offer is, say, $71 on a $20 stake. The insured refund if the next leg fails is $20 in site credit worth about $12.20 after conversion. The bettor is being asked to hold a $71 position with a $12.20 floor. That's the trade. It's not a bad trade — cash-out offers are themselves priced with a house edge, and refusing them is often correct — but it's a trade with a number attached, and the number is smaller than the word "insurance" implies.
The word does a lot of work
"Insurance" in a casino context is a marketing term, not a financial one. There's no underwriter, no risk pool, no solvency requirement. The book isn't transferring your risk to a third party; it's offering you a conditional rebate on a bet it expects to win. That's a legitimate product. It's also a product whose perceived value depends heavily on the label. When the same offer was framed as "stake back if one leg fails" instead of "insurance," attach rates in a 2023 A/B test at one of the four books dropped 14% — smaller than the 26% from the payout display, but pointing the same direction.
The 26% figure matters less as a prediction than as a proof of concept. Bettors will change behavior when you show them the arithmetic. They will not change behavior when you show them more words.
Where this leaves the promo economy
Sportsbook promos in the U.S. have been on a multi-year slide from cash-equivalent to credit-based, and insurance tokens are the leading edge of that shift. A $200 risk-free bet in 2019 was a genuine $200 of value to a new customer. A 2025 "no sweat" token is a conditional rebate on a capped stake, paid in expiring credit, worth somewhere between 40% and 65% of face value depending on how the bettor uses it. The industry didn't hide this, exactly. It just stopped saying it out loud.
The 26% number suggests the information asymmetry has a shelf life. Once a book ships a payout breakdown, attach rates don't recover — the informed cohort in the sample stayed below 32% for the full 11 months after the feature launched. Bettors don't forget a subtraction problem. They might still take the insurance, but they take it knowing what it's worth.
That has a second-order consequence worth watching. If insurance attach falls, the promo's retention value falls with it, and books will either raise the refund cap (expensive), improve the conversion terms (also expensive), or shift the acquisition spend to a different mechanic. The most likely replacement is already visible in the market: boosted parlays with no refund at all, where the value is baked into the price rather than deferred into credit. Those are harder to evaluate at mid-leg, not easier. A bettor who learned to subtract refund value from cash-out value will find nothing to subtract on a boosted parlay — just a number that looks better than the market and a set of legs that still has to hit.
So the open question isn't whether bettors will keep insuring parlays. It's whether the payout-breakdown feature spreads. If it does, the 26% decline becomes the industry's new baseline and the promo arms race moves somewhere less legible. If it doesn't — if breakdowns stay a niche tool that a minority of bettors open once and forget — then the number stays a data point in a study rather than a change in how the market prices risk. Either way, the bettor who runs the subtraction is playing a different game than the one the promo was designed for. That's usually the tell.