How to Tell Which Sportsbooks Are Sharp
Judge a book by whether the market moves toward it, not by how big it is.
In a test of 27,000 games and more than 250,000 player props, every major book looked accurate at the close, so the useful test became which books disagree early and get followed, and those books deserve the most weight in your fair price.
Weight your fair odds toward the books the market follows, cap the books that copy and limit hard, and check that your weights fit the sport you bet.
Every +EV number depends on the books behind the fair price
Every +EV bet is measured against a fair price, usually a no-vig price (the odds with the book's cut removed) blended from several books. If that blend leans on the wrong books, the edge you think you have may not be there. In the Optimizer, the EV column is simply your price compared with the fair price, so a wrong fair price makes every EV on the board wrong.
The video tests which books deserve that trust using closing lines from 13 or more sportsbooks and exchanges, more than 1.1 million line moves and 22.5 million exchange trades, all graded against box scores. A book earns trust in two ways: its closing prices have to match how often things actually happen, and the rest of the market has to move toward it.
Why closing-line accuracy cannot pick the sharpest book
The first test removed the vig from every closing line with the probit method and compared each fair price with the result, and the close was well calibrated: when DraftKings closed a side at 75%, it won 74% of the time. On 9,907 games where the line moved three or more points, the side the money moved toward went from 51.2% at the open to 57.5% at the close and won 58% of the time. That gap is closing line value (CLV), and it is why beating the close is such a strong sign of a good price.
The catch is that every major book passed the accuracy test, even Hard Rock. FanDuel and DraftKings were basically identical at the close across 12,883 games, because by game time the big books sit within about a point of each other after copying one another and adjusting to sharp money. Accuracy tells you the close is good, but it cannot tell you who got there first.
The market-follow test: who disagrees early and gets followed
To find who leads, the video compared each book's price 24 hours, 6 hours and 1 hour before the start with everyone else's, then measured how much of any disagreement the market closed toward that book by game time. If one book says 55%, the rest say 50% and everyone ends at 54%, the market closed 80% of the gap toward that book. That ran on 6,087 games with timestamped line history, plus actual trades from Kalshi, ProphetX and Novig.
The biggest books did not lead. The market moved only about 5% of the way toward DraftKings when it disagreed, but 68% of the way toward Novig. The video's explanation is that on an exchange you trade against other bettors, many of them sharp, so the price moves when sharp money hits it, while a sportsbook that limits sharp bettors mostly reacts to everyone else.
Share of the gap the market closed toward each book (game lines)
Exchange: the clearest leader in the test.
Exchange.
One of the biggest books in the country.
These books followed the market rather than moving it.
Player props: same player, same line, whose price was right
Graded one at a time, every book looked fine on props too, so the video put books head to head on the same player and the same number: 252,387 closing prop prices and 417,833 matchups. The measure was how much paper edge showed up in results. If a book's close says a bet hits 52% and you got a price that only needed 48%, the question is how many of those four points actually came through.
Beating Novig's close was worth 121% of the paper edge, followed by Kalshi at 73%, FanDuel at 68%, DraftKings and Bet365 at 47%, BetRivers at 43% and Hard Rock at 31%. A result above 100% means beating that close was worth even more than the close itself suggested. Sportsbook prop overs also carry a built-in markup the video calls the over tax, which has its own lesson.
Bet365 was the worst book tested on NFL props (5%) and the best on MLB props (73%), while DraftKings went the other way (93% on NFL props, 39% on MLB). One set of weights for every sport leaves value on the table. These results describe past data, and books change how they price markets.
Turn the results into fair-odds weights
The goal is a blend of books whose fair price lands closest to the closing number, where a book earns more weight for being early and right and loses weight for repeating what another book already said. The video caps Hard Rock, BetMGM, BetRivers and Fanatics at 0.1 because they limit hard, so they cannot steer the fair odds.
Only the ratios between weights matter: a book at 7.5 next to a book at 1 counts seven and a half times as much. Pinnacle is in the data, with 6,008 closing lines, and ties FanDuel and DraftKings at the close, but without minute-by-minute Pinnacle history the video does not rank it on who moves first.
The weights shown in the video
- Game lines: Novig 7.5, Polymarket 3, FanDuel and Kalshi 1.1, DraftKings 1.
- Player props: Novig 7 and Kalshi 7, then DraftKings, ProphetX and Pinnacle 1.4, and FanDuel 1.
- Books that limit hard (Hard Rock, BetMGM, BetRivers, Fanatics): capped at 0.1.
Set the weights where your EV is calculated
In the Optimizer, every row is a bet at the book you picked up top, shown beside its fair price. The default All books setting counts every book the same, so Hard Rock gets the same say as Novig, and changing the weights changes the fair price behind every row and every EV number you see.
Book size is not sharpness. Use the weights as a starting point, check that they fit the sport and market you bet, and still confirm the live price at your book before you decide.
Set fair-value weights in the Optimizer
Open Weights
In the Optimizer, click Filters, then click Weights next to fair value.
Choose a preset
All books is the default and weights every book equally. Under Recommended, pick Matt's Game Line Weights or Matt's Prop Weights to fill in the weights from the video, then click Apply.
Save your own
If you build your own blend, save it as a custom preset so you can reuse it.
Article FAQs
What makes a sportsbook sharp?
A sharp book's closing prices match real outcomes, and the rest of the market moves toward it when it disagrees. In the video's data nearly every book passed the first test, so the second one is what separates them.
Is the biggest sportsbook the sharpest?
Not in this data. On game lines the market moved about 5% of the way toward DraftKings and 68% of the way toward Novig, so book size and sharpness are different things.
Why do weights only matter as ratios?
Fair odds are a weighted blend, so what counts is how much each book counts relative to the others. A book at 7.5 counts seven and a half times as much as a book at 1.
Where do Pinnacle and ProphetX fit?
Pinnacle ties FanDuel and DraftKings at the close in the video's data but is not ranked on who moves first because minute-by-minute history was not available. In the prop weights shown, Pinnacle and ProphetX each get 1.4.
Set your fair-value weights
Open the Optimizer, choose a weight preset under Filters, and see how the fair price behind every row changes.