Bankroll Rules for Same Game Parlays

A same game parlay should never get the stake you'd put on a straight prop, because the sportsbook takes a far bigger cut of it. It's an easy thing to miss. The bet slip shows one ticket, one stake and one price, so it feels like one bet. It isn't. You're paying for several bets at once, and the book has already priced out the connection between them that made the combination look attractive to you in the first place.

This matters most to prop bettors, because props are what SGPs get built from. A points over, a rebounds over, a team total. If you size that ticket like your usual unit, you're making your most expensive bet at your normal stake.

Start with a plain -110 bet. Its implied probability is 52.4%, and on a true coin flip it costs you about 4.5% of your stake over the long run. String independent legs together and the cost compounds. Jeff's Edge estimates the effective hold at roughly 9% for two legs, 13% for three, 17% for four, nearly 21% for five and 31% for eight.

Those are ordinary parlay figures. Same game parlays add a second layer. When legs come from the same game they move together: a quarterback throwing for 275 yards makes his team more likely to win and the total more likely to go over. Wizard of Odds shows an illustrative NFL correlation matrix where those pairings run from 0.28 to 0.42. The book models that correlation, cuts the payout to match, and then takes a margin on top.

The Wizard's full example lays it out. Legs with true chances of 56%, 48% and 52% hit together 14.0% of the time if they're independent. Correlation lifts that to 18.9%, which makes the fair price about +429. The book offers +350, which implies 22.2%. That's a 14.9% house edge, and the article puts SGPs generally at 15% to 25% or more, three to five times the edge on a single bet.

The stake comparison is the bit to take away. Three separate $10 singles at -110 on 52% legs have an expected loss of about $0.22. One $10 SGP at +350 with a true 18.9% chance has an expected loss of about $1.49. Same money, roughly seven times the cost.

Leg count. Every leg you add cuts your chance of cashing and raises the hold. With 50/50 legs and no correlation, the chance of hitting halves with each leg. A four-leg ticket hits about 1 time in 16. The book's price lags further behind, too: Jeff's Edge cites a four-leg 50/50 parlay that should pay +1500 but gets offered around +1228. On a $100 stake that's about $270 short.

Hit rates and fair payouts here are straight arithmetic on independent coin flips. The hold column comes from Jeff's Edge, which also says a five-leg SGP's edge can go above 35%. None of my sources gives a hold figure for ten legs.

Correlation priced into props. Establish The Run pulled a DraftKings NBA example. Brandon Miller under 17.5 points paired with Jarrett Allen over 2.5 assists had a fair uncorrelated price of +267, and DraftKings offered +258. Pair the same Miller leg with Allen's under and the fair price was +224 against an offer of +208. The book shaded both directions, so the discount was there whichever side you picked.

Leg caps and void rules. Published leg caps don't agree with each other. TopEndSports, last updated June 30, 2026, lists DraftKings at 12 legs and FanDuel at "10+", and gives no figure for BetMGM. Other pages quote higher numbers, so check the cap in the app. The void rules matter more for your bankroll anyway. According to Action Network, FanDuel recalculates the price on the remaining legs when one leg voids. DraftKings, BetMGM and Caesars void the whole ticket. ESPN BET's terms let a push still leave a losing ticket as a loss. On a long ticket, a push can wipe out four winning legs.

The SGP gets lumped in with two other bets. It's priced differently from both, so it shouldn't share a stake rule with either.

General bankroll advice goes up to 1% to 5% a play, and Action Network's Josh Appelbaum calls 3% "a good medium". That's advice for flat straight bets, and he says outright that flat betting means skipping parlays. The only specific SGP rule I can find in print is from The Spread: size parlays and SGPs at 0.25 to 0.5 units, a "fun size" stake, with the unit itself at 0.5% to 2% of bankroll. I'd go further and tie the fraction to leg count. Use half a unit for two or three legs. Use a quarter unit for four or more, or for any ticket where the legs are obviously correlated, because that's where the book has taken the most out. Your base unit is set in the step-by-step bankroll framework, and the SGP stake should sit inside it, never next to it.

The strongest argument on the other side comes from Jeff's Edge. He still bets a "game-script SGP", built around one expected game flow, on the view that a sharp read can beat a book's correlation model. He also notes that books model correlation differently, so the same combination can vary 10% to 20% in price between them. That's a real argument for shopping and for betting some SGPs. It isn't an argument for sizing them up. Unless you can show the edge in EV terms, the ticket gets the fun stake.

Stacking favorites. Bundling short prices you'd bet anyway turns cheap singles into one expensive ticket. Jeff's advice is to bet those individually so the vig doesn't compound.

Stacking longshots. Every extra plus-money leg adds hold on top of an already tiny hit rate. DeucesCracked recommends two or three legs, and another of its guides says three or four.

Adding a "lock". A -400 leg barely moves the payout, but it still has to win and it still carries margin. It also fools you into thinking the ticket is safer, so you stake more.

Chasing with bigger tickets. Raising the stake or the leg count after a miss is the fastest version of what Establish The Run calls "perhaps the quickest way to go broke."

Two checks help. First, The Spread caps exposure at 2 to 3 units per game, SGPs included. A side, a prop and an SGP on one game are, in Jason Turner's phrase, "one opinion wearing four different outfits." Second, promos. Profit boosts, odds boosts, insurance and no-sweat tokens make a ticket cheaper, but they don't make it bigger. Apply the boost to the stake from the table and don't raise the stake to get more out of the boost. None of my sources puts a number on how much a given boost recovers, so do that math yourself before treating a boosted SGP as positive EV.

Mixing SGP results into your prop record hides two things. The SGP losses look smaller than they are, and any real prop edge looks weaker. Log SGPs as their own market with stake, price and result, and set them a separate budget, as DeucesCracked advises. Then compare ROI by bet type every month. A prop tracker with closing line value can split the record by market so the two numbers never blur. If the SGP line is running near the -15% the Wizard's math predicts, the ledger is working. It's telling you what the entertainment costs.