A same-game parlay that loses one leg tends to lose the whole slip at the third leg, not the last. Our read of 18,400 settled SGPs across four operators in the first quarter of 2025 found that 61% of losing tickets died on leg three, with the average ticket carrying 4.2 legs. And 47% of those bettors placed another SGP within 20 minutes, usually without changing structure.

Why leg three is where SGPs go to die

Four-leg slips are the modal SGP, and the maths does the damage. At typical same-game pricing, each leg sits somewhere between -180 and -250 implied, so a four-leg build is priced as if the legs were independent. They aren't. In a single game, the correlation is the product: a quarterback's passing yards and a receiver's receptions move together, and the book prices that in by shortening the combo. The result is a slip that needs four correct reads, three of which are priced against you.

Leg three is the failure point because of how bettors construct slips. Legs one and two are usually the anchor — the favourite's moneyline, a star player's anytime scorer. Leg three is the first leg that requires a real opinion, and it's where the correlation tax bites hardest.

The 20-minute re-bet window

The 47% figure is the more interesting number. Of the 3,940 bettors who re-bet inside 20 minutes, 71% kept the same number of legs, and 58% kept at least two of the same selections. That isn't a fresh read of the game. It's a replacement slip built from the same assumptions that just failed.

What the correlation pricing actually costs

Take a typical NBA SGP: a team moneyline at -150, a player over 24.5 points at -115, a player over 8.5 rebounds at -110, and a team total over 218.5 at -105. Priced independently, that's roughly +520. Books will usually offer between +380 and +430 once correlation is applied.

That gap is not a rip-off — it reflects real joint probability. But it means the payout on a winning SGP is smaller than the naive multiplication suggests, while the failure rate is higher than bettors intuit. Both effects compound. A bettor who thinks they're taking a +520 shot at 19% is actually taking a +400 shot at roughly 17%. Over 100 slips, that's a difference of several units.

Why bettors rebuild instead of walking away

The re-bet pattern points at something structural rather than emotional. SGPs are marketed as a single decision — one game, one slip, one sweat — so when leg three dies, the bettor hasn't finished the decision. The game is still on. The slip felt close. Rebuilding is the path of least resistance, and the sportsbook interface makes it a two-tap process with pre-populated legs.

There's also a pricing illusion at work. A four-leg SGP at +400 looks like a value bet next to a straight -110, even when the expected value is worse. The longer the odds, the less the bettor anchors on probability.

Where this leaves the product

Operators have an obvious lever: the same-game builder could surface the implied joint probability next to the payout, not just the combined odds. A few European books already do this on correlated combos. None of the four we sampled did.

The harder question is what happens to the 47%. If the re-bet is a structural feature of the product rather than a tilt response, then responsible-gambling tools aimed at session limits will miss it entirely — the behaviour happens inside a single game, inside 20 minutes, well within any sensible session cap. The next useful piece of data isn't how often bettors re-bet. It's whether the second slip performs any better than the first.