The Lab · Reading the market · updated 2026-09-05

Which sportsbook has the best odds?

We compared eight sportsbooks across 1,440 moneyline markets on our own recorded closing prices. The book that wins most only posts the best number 27% of the time — which changes the question from which book to use to how many to have.

It's the first question anyone asks, and it has a real answer sitting in our database: we record every book's price on every game we grade, so we can just count.

Across 852 graded MLB games and 1,440 priced moneyline markets, here is who posted the best number.

The table

Every book quoted essentially the same slate, so nobody here is ranked high for showing up rarely. Value is the implied cost you save by having that book available versus taking the next-best price, averaged over every market it quoted. Gain when best is how much it wins by on the markets it does win.

The number that actually answers the question

26.8%.

That's how often the top-ranked book posted the best available price. Which means that if you bet exclusively at the best book in this table, you took a worse number than was available on roughly three of every four bets.

No book is "the book with the best odds." The best book is the best book about a quarter of the time.

Look at how tight the top of that table is: Caesars 26.8%, FanDuel 24.1%, Hard Rock 24.0%, DraftKings 23.2%. Four books within four percentage points of each other. And Caesars and BetRivers tie exactly on Value at 0.10pt — the ordering between #1 and #2 isn't a real distinction, it's rounding.

Two questions that look like one

The Value column is easy to misread, and we misread it ourselves before writing this.

It counts a book's gain only on the markets it won, then divides by every market it quoted. So it measures the marginal worth of having that book on your list — what you'd give up by deleting it and taking the next-best price instead.

That is a different question from which single book should I bet at, and the answers can genuinely invert. A book that wins rarely but by a mile is enormously valuable to someone comparing prices and worth nothing to someone who never compares. A book that never posts the best number but is never more than a whisker behind scores 0.00 on Value — reading as "worthless" — while being the cheapest place to be loyal to.

BetRivers is the visible case here: it wins less often than four books below it (19.9%), but when it wins it wins biggest (0.50pt), which is exactly the profile that matters to a shopper and not at all to someone with one account.

A detail that gets missed: ties are normal

Add up the "times best" column and you get 2,032 across 1,440 markets. That's not an error — about 1.4 books share the top price on a typical market.

Best prices are usually ties, not victories. Most of the time several books are level and it makes no difference which you use; the money is in the minority of markets where one book is genuinely out on its own.

What is this worth in actual money?

Implied probability points don't read as anything, so convert them.

Expected value per unit staked is (your probability − break-even) × (1 + net odds). At even money, 1 + net odds is 2 — so one implied point is worth about 2% of your stake in EV. It's about 1.5% on a −200 favourite and 3% on a +200 dog.

So the whole spread in that table, 0.10pt at the top to 0.01pt at the bottom, is 0.09 points — under 20 cents per $100 wagered. Depending on your volume, the difference between the best book and the worst is somewhere in the low tens of dollars across a full season.

That's real. It is also not the thing standing between you and a profitable season.

The comparison that puts it in scale

The house edge on a typical two-way moneyline runs about 1.75 points of implied probability. The best book in our table is worth 0.10 points.

Shopping recovers roughly 6% of the vig. Worth having. Not salvation, and not remotely close to it. Anyone selling you a book recommendation as an edge is selling you six percent of the house's cut and calling it an advantage.

So what should you actually do?

Shop, don't choose.

The value of comparing prices is exactly the cost of not comparing them. Take the best available number every time and you keep it; stay loyal to one book and you pay it, every bet, forever.

Practically, that means:

• Two accounts beat picking the right one. Going from one book to two captures most of the available spread, because ties are common and the top four are closely matched. Going from four to eight adds much less.
• Which two matters less than you'd think. Any two of the top five are fine. Don't move accounts over three hundredths of an implied point.
• Check the number, not the brand. A book's ranking here is an average over 1,440 markets. On the game in front of you it tells you nothing — only the actual posted price does.

Our daily board shows the best price across every book we track on every game, which is the whole point: it's the comparison, done for you, before you bet.

What this doesn't tell you

Honest limits, because they're real ones:

• MLB moneylines only. Book-to-book behaviour differs by sport and market. Nothing here transfers to NFL spreads or player props, and props in particular are where books differ most.
• Closing prices only. We compare the last pre-game number each book posted. A book that's sharp at open and slow at close, or the reverse, isn't distinguished here.
• It says nothing about limits, holds, or getting restricted — which for a winning bettor eventually matters far more than a tenth of a point.
• This is a ranking, not an endorsement. The order comes out of a query over prices we recorded ourselves; nobody buys a position in it. How this site makes money is on the disclosure page.

The reason this article is short on drama is that the finding is genuinely small. That's the finding. The interesting part isn't which book won — it's that the winner only wins a quarter of the time, which quietly turns "which book?" into the wrong question.

← Back to The Lab · How the model works · The record