🎲 Why Most Parlays Destroy Your Bankroll

Parlays account for under a third of the money Americans bet on sports and almost 60% of what sportsbooks keep. That gap is the most important number in this article, and almost nobody who bets one has seen it.

It does not mean parlays are rigged. Nothing is being done to you that is not printed on the ticket. It means the arithmetic behind a parlay works differently from the arithmetic behind a single bet, in a direction that gets worse with every leg you add, and it punishes one particular decision much harder than the others.

I have been betting on sports for more than thirty years, and I rarely play parlays, because I can do this arithmetic. That is not the same as saying every parlay is unplayable, and I do sell one, built the opposite way round from the tickets the books put in front of you.

Those two things belong together, and that's the most reliable instinct I have. When I find something the sportsbooks are conspicuously uninterested in me betting on, I know I am onto something. They advertise same-game parlays relentlessly. They do not advertise two-leg tickets built out of heavy favorites. That asymmetry is not an accident, and by the end of this article you will be able to read it off a table.

🎲 Every parlay I have ever released is posted with its result attached. See the full archive.

💰 Under a third of the money, almost 60% of the revenue

New Jersey publishes its betting numbers in detail, which makes this checkable rather than something I am asking you to take on faith.

In 2024, bettors put $12.812 billion through New Jersey's sportsbooks on completed events. Parlays were $3.671 billion of that, or 28.6%. Those same parlays accounted for $641 million of the state's $1.082 billion in sportsbook revenue, 59.3%. Under a third of the money went into parlays, and almost 60% of what the books kept came back out of them.

The mechanism is the hold, which is the share of every dollar wagered that a book keeps once all bets are settled. New Jersey's parlay hold in 2024 was 17.5%. On everything else, it was 4.8%. A parlay is not sold to you at a worse rate per leg. It charges you the ordinary rate again on every leg, and the rest of this article is what that does to a ticket.

This is not a New Jersey quirk, and it is getting worse rather than better. Across Illinois, New Jersey, Maryland, Colorado, Oregon and Mississippi, measured over the same May-to-August stretch in consecutive years, the parlay share of sportsbook revenue went from 51.8% in 2024 to 59.8% in 2025.

This is understood well enough that it has moved money outside sports betting entirely. Jim Chanos, the short-seller who famously bet against Enron before it collapsed in 2001, began shorting DraftKings in May 2021 because he thought the business model was weak. He covered that short in July 2022 and booked a profit on it anyway. Explaining the reversal to the Financial Times in December 2023, he said the thing he and his firm had underestimated was "what bad bettors the US gamblers are," pointing at the growth in the harder-to-price wagers where the odds are least transparent, parlays among them.

He was not describing bad luck. He was describing the table below.

✖️ Does the house edge add up, or multiply?

Here is the part that catches people.

A parlay pays the product of its legs. Two legs at even money do not pay you twice; they pay you roughly four times, because the second price is applied on top of the first. That is what makes the payout attractive. It is also what makes the cost behave the way it does, because the book's margin rides along on exactly the same multiplication.

So a four-leg parlay does not cost you the house edge four times over. It costs you the house edge compounded four times over, and compounding is a great deal less forgiving than addition.

“A parlay does not add the house edge four times. It multiplies it four times.”

The practical consequence is that the price you start each leg at matters more than anything else you decide about the ticket, and it matters more with every leg you add.

📉 What each extra leg actually costs

I keep a database of closing prices covering 405,822 team games across seven sports, dating back to 2003. It is the thing I built the company on, and it lets me ask a question most bettors never get to ask: what does a given kind of bet return if you simply back it every time, forever, and never think about it at all?

That is what I mean by a blind bet. Backing one side every single time it came up, with no handicapping applied and no opinion about the two teams. It is the starting position before anybody does any work, and comparing two bets blind is the only fair way to see which one asks more of you.

Two moneyline bands sit at opposite ends of that database. Heavy favorites, priced at −300 or shorter, returned −1.58% blind across 59,323 examples. Longshots, priced at +300 or longer, returned −9.03% blind once I set college basketball aside, and I will come back to why that exclusion matters.

Run both through a parlay:

One longshot costs about seven and a half points more than one heavy favorite. Four of them cost about twenty-five points more. The mistake compounds faster than the payout does, and it is the same mistake in every row.

Now the exclusion. Include college basketball and the longshot column gets considerably worse, because that sport carries 32,209 longshot games at −25.04%, the steepest price I have measured anywhere. On the full seven-sport pool, a longshot returns −19.73% blind, and four longshot legs come to −58.48%. I have left college basketball out of the table because I would rather show you the version that understates my case. If you are stacking college basketball longshots, the real number is roughly double what you see above.

🧮 The decision that does the damage

Notice what is missing from that table. It says nothing about which teams are playing, which sport they are playing in, or how well you handicap them. The entire spread between those two columns comes from one decision made before any thinking starts, which is what price you chose to build from.

That is why the same-game parlay gets pushed at you so hard. The individual legs are priced normally enough. What the format does is put four longer prices onto one ticket efficiently, and four longer prices is the bottom-right corner of that table.

No blind parlay is a good bet at any leg count. Five legs of favorites still return −7.65%, against −1.58% for a single one. If losing the least is the whole goal, the answer is not a better parlay. It is one bet.

What the table settles is a narrower question, and it is the one an actual parlay bettor is asking: given that you are going to play one, where does it cost the least? That answer is unambiguous, and it is short prices and few legs.

🎯 How I actually play it

The damage does not happen the way most people picture it, so start there.

No single parlay slip ruins anybody. What ruins a bankroll is a negative expectation applied to ticket after ticket over the course of a season. Someone who plays four-leg longshot parlays as a habit is running a −31.52% expected return through his account every week, and there is no bet size small enough to make that arithmetic work over a year. A season of two-leg favorite tickets at −3.14% bleeds instead, slowly, which is survivable and still not free. What separates those two bettors is the two decisions above, made before either of them handicapped a game.

The arithmetic points at a few practical things, and the one people skip comes first.

Decide honestly which kind of bettor you are on this ticket. A small parlay played for entertainment, at a stake you would not miss, is a perfectly reasonable thing to buy, and I put one on occasionally myself. What it is not is a strategy. The trouble starts when a bettor treats an entertainment purchase as an investment and sizes it accordingly, which is also how most losing seasons unfold. My bankroll rules apply to a parlay slip the same as anything else, and a bet with a low chance of cashing belongs at the bottom of that ladder rather than the top.

Keep the leg count down. The difference between two legs and five is the difference between −3.14% and −7.65% before you have handicapped a thing, and every leg you add is another independent way to lose the whole ticket.

Build from the short end of the board. Everything above says the starting price is the decision that compounds, so the lowest available starting price is where a parlay should be built from. That is moneyline favorites rather than the long prices the books put in front of you.

That last point is the whole design of my Parlay Club, which builds two- and three-leg moneyline tickets from favorites and nothing else, occasionally stretching to five legs. It is the same argument as this article, applied. It is also the opposite of what a same-game parlay promotion is built to sell you, which is rather the point.

📊 Where this sits against everything else

None of this says parlays cannot be beaten. It says the blind version of the bet starts you well behind, and that how far behind is your choice rather than the book's.

The gap between a blind bet and a good one is the edge a handicapper actually adds, and my honest estimate of what a strong professional is worth is two to five extra wins per hundred over a blind bettor in the same spot. Against a two-leg favorite ticket at −3.14%, that is a fight worth having. Against a five-leg longshot ticket at −37.70%, it is not close, and no amount of skill closes it.

It is also worth knowing that a losing run inside a winning approach is normal rather than a signal, which is what variance does to a record over a season. Parlays make that swing wider, because a ticket either cashes whole or dies whole.

I have made 89,386 official picks since 2001, and every one of them is posted with its result attached, parlays included. You are welcome to check the ones that lost.

If you want my parlay card for the season, the Parlay Club is here. It runs through the football and basketball calendar and takes the summer off. Either way, the arithmetic above does not change.

Blind-bet figures come from my own database of closing prices across 405,822 team-games in seven sports, 2003 through 2026, and are what a bet in that situation returned rather than what any pick of mine returned. Parlay returns are calculated from those figures, assuming the legs are independent and the ticket pays the product of their prices, which is how a standard parlay is priced. Correlated legs behave differently, and some books price that correlation. New Jersey handle, revenue, and hold figures are from the state's official 2024 annual gaming report. The six-state revenue shares cover Illinois, New Jersey, Maryland, Colorado, Oregon, and Mississippi over May to August in each year, a summer window with little football in it, so it reads lower than a full season would. Lines move, so confirm the current price before you bet it.