DraftKings reported $3.67 billion in revenue last year. A meaningful slice of that came from parlay bets, not because parlays are a good product for bettors, but because they are an extraordinarily good product for books, and most bettors have no idea how badly the math is working against them every time they build one.
The gap between what a parlay should pay and what it actually pays is not a rounding error. It is the business model.
A two-team parlay at -110 on each leg hits roughly 27.8% of the time at true probability with no juice. At fair odds, that ticket should pay around +260, meaning a $100 bet returns $360 total. Most major US books pay +260 to +264 on a standard two-teamer. Some pay as low as +200.
That spread, from +260 down to +200, represents a house edge of roughly 10% on what looks like a simple two-leg ticket. For context, American roulette runs about 5.3%. Standard slot machines run 8% to 12%. Recreational bettors who would never sit at a slot machine for an hour build four-team parlays every Sunday morning without a second thought about the math underneath them.
The house edge doesn’t stay at 10% as legs get added. It compounds. A four-team parlay at standard -110 pricing carries a house edge north of 30%. A six-teamer pushes past 40%. Each additional leg isn’t adding proportional risk. It’s multiplying the book’s take on every single component simultaneously, and the payout never keeps pace.
A $25 five-team parlay paying $800 feels like asymmetric upside, a trivial amount at risk against a genuinely exciting return. That framing is not accidental. It’s the entire architecture behind every “Parlay of the Day” push notification, every jackpot teaser running on DraftKings and FanDuel from September through February, every social media post showing a screenshot of a $38 ticket turning into $1,200. The product is engineered to be evaluated by what you could win, not by what the ticket actually costs relative to the true probability of winning it.
Here is the actual math on that $25 five-teamer. At true fair odds, a five-leg parlay at -110 each should pay roughly +2300, meaning a $25 bet returns around $600. Standard US books pay somewhere in the +2000 to +2100 range on a five-teamer. The shortfall on any single ticket isn’t dramatic. Across the millions of parlay tickets processed every NFL Sunday, it is a massive, systematic transfer from bettors to operators, one $25 ticket at a time
The appeal of an SGP is correlation. A bettor who thinks Jalen Hurts is going to have a big game and the Eagles are going to cover feels like those outcomes belong together on one ticket, because they are related. If Hurts throws for 310 yards and two touchdowns, Philadelphia almost certainly had a productive offensive day, which is closely connected to whether they covered the spread. That relationship is real.
Books know it too, and they price SGPs as if it isn’t there. Each leg gets priced independently, as if the outcomes share no relationship, and the combined payout reflects none of the correlation that makes them more likely to land together. FanDuel and DraftKings began pushing SGPs aggressively around 2020 and 2021, and analysts tracking operator margins since then have consistently found SGP hold rates running well above standard parlay hold rates. [VERIFY: specific hold rate figures from public operator disclosures or earnings calls] The margins are wide enough that multiple sports betting researchers have described SGPs as the highest-edge product on a modern retail sportsbook menu.
That doesn’t mean every SGP is a bad bet in every circumstance. It means the book has almost certainly priced the correlation away before the ticket is ever displayed, and confirming that requires calculating fair value on each leg and the combined probability yourself, which almost no recreational bettor does.
There is a specific, documented condition under which building a parlay is defensible from an expected-value standpoint. The legs need to be correlated in a way the book has failed to price in, and you need to be able to verify the discrepancy before placing the bet.
The most cited example among sharp bettors involves game totals and team totals. If a book posts a full-game total of 47.5 and the two individual team totals add up to 49, a mathematical inconsistency exists somewhere in the market. A bettor who identifies which side of which total is mispriced and combines them into a two-leg parlay may be capturing genuine positive expected value, because the correlation between the game total and its components is direct and the book has priced them in a way that doesn’t reconcile.
This kind of opportunity exists. It is not common. It requires simultaneous access to multiple markets, the ability to calculate implied probabilities from each line quickly, and the speed to act before the book corrects the gap. It is not what most bettors are doing when they build a Sunday morning parlay around their three strongest feelings of the week.
Assume a bettor has done the work and identified a genuine positive-EV four-team parlay. The math confirms the book is offering better than fair value. That’s legitimately good.
A +EV four-teamer still loses around 85% of the time. The payout has to be large enough to overcome that losing frequency over a significant sample, and the bettor needs both the bankroll depth and the psychological steadiness to absorb a long string of losses before the edge shows up in the numbers. Most people who believe they’re playing +EV parlays are not, but even the ones who are face a variance problem that makes straight betting look stable by comparison. You can be right about the math and broke by November if the sizing isn’t proportional to the actual hit rate.
Sharp bettors who do play parlays in specific, well-defined spots keep the stakes small relative to total bankroll. Sizing a four-team ticket at 5% of your roll because the payout looks compelling is still a sizing mistake even if the ticket carries positive expected value. The bet has to reflect the probability of winning it, not the size of what winning it would feel like.
There is one genuinely defensible case for recreational parlay betting, and it requires being direct about what the money is actually for.
A bettor who sets aside $40 a week as entertainment spending, builds two small parlays for the experience of having action across multiple games, and treats the whole thing as the cost of a more engaged Sunday afternoon is making a rational decision. The $40 is a budget for fun. The book’s edge on the ticket is the price of the entertainment, and plenty of people spend more than that on worse ways to spend four hours.
The problem is that most recreational bettors are not operating this way. They’re building parlays with the genuine belief that their combination of picks is clever, that this week’s five-teamer is different from last week’s losing ticket, that they’ve identified something the book missed. That belief is precisely what a 30% house edge is designed to feed on, and it’s remarkably durable even after repeated losses.
Before building any parlay, the math that exposes the book’s edge takes about two minutes on a phone.
Convert each leg’s American odds to implied probability. A -110 line implies 52.4%. Multiply the win probabilities of every leg together. That product is the combined win probability. Divide 1 by that number to get the fair decimal odds, then convert to American odds and compare directly to what the book is offering on the ticket.
On a standard two-teamer at -110 each, fair value is roughly +260. If your book is offering +264, you’re losing about 8% to the house. If they’re offering +230, you’re losing closer to 17%. On a four-teamer, fair value is around +1228. Most books pay somewhere between +1000 and +1100. That gap is yours to give away or keep.
Do the calculation before the next ticket. Whatever you decide after running the numbers will at least be a real decision rather than a guess dressed up as confidence.