How do you research a player prop before betting it?

Most guides to this start with the player. Ours starts with the price, and the reason is boring: you can spend twenty minutes deciding you like a hitter and then find the number on offer was never worth taking. Doing it the other way round removes most of the board in seconds.

This is the sequence our own board runs every morning, in the order the code runs it.

1. Build a fair, before you look at anything else

A sportsbook price has the margin baked in. The first job is to get an estimate of the number without it.

We build that from exchanges and sharp shops, which are the books whose prices sit closest to the real number: NoVig, ProphetX, Pinnacle, Polymarket, Circa, BetOpenly. Their prices are treated as a measurement of the truth rather than as offers.

If you are doing this by hand, the shortcut is to take both sides of the line from one book and strip the margin out yourself. How to devig odds walks through that, and the no vig calculator does it for you in a few seconds.

One market where this does not work: anytime touchdown, and any other one-sided market. There is no under, so there is nothing to devig, and the only comparison available is against other books.

2. Never bet into the book that built your fair

This is the rule with the most consequences and the one you will not find in other guides.

If a sharp book’s price is what you are using to decide the true number, then that book’s price can never look like value to you. It is your yardstick. Measuring the yardstick against itself gives you zero, always, by construction.

There is a more uncomfortable version. The same books that make the best fair also tend to be the best price on the board. On our own board, on one night, the cheapest book on home run props by a wide margin was an exchange charging about 2% where most sportsbooks charged 6.5% to 9%. It is genuinely the best number available. It is also the number you are using as truth. You can use it as a price or as a yardstick, not as both.

So our board classifies every book into one of three kinds, and the kind decides what a price means:

Whatever your own list looks like, keep the two roles apart.

3. Form, with a window and a floor

Only now does the player come into it.

Our form number is 0.6 times the last 10 games plus 0.4 times the season. Recent is weighted higher because it carries newer information; season is in there because ten games of a yes-or-no outcome is a very small number and on its own it is mostly noise.

Under 20 games of history we do not show a rate at all, and thin samples get dropped rather than displayed small. A number on screen gets read as a fact regardless of what is under it.

That floor bites unevenly and it is worth knowing which side you are on. On our board the median batter has 109 games of history and the median pitcher has 27, so the rule excludes about 5% of batters and 30% of pitchers. A strikeout prop’s hit rate rests on a quarter of the history behind a hits prop’s.

More on the weighting and the traps in what stats player prop tools use.

4. The matchup, as numbers with ranks on them

The matchup step is where most research goes wrong, because it is where adjectives live. “Good matchup” and “soft defence” are not inputs.

What makes it usable is a rank and a season attached to every claim. Not “this defence struggles against tight ends” but “this defence allowed the most red zone targets to tight ends in the league last season, 2.18 a game against a league low of 0.71.” The second one you can check and disagree with.

The market-specific inputs are different per market and they have their own pages: home runs, strikeouts, total bases, stolen bases and walks. For football, what to check for anytime touchdown props.

For a batter, the matchup also takes in his record against tonight’s pitcher, and why that sample is rarely big enough to lean on is in does batter vs pitcher history matter.

5. Place it at a number you can actually get

The last step sounds trivial and is where a lot of research quietly evaporates.

A prop is only worth what you can take it at. Our board prices to the books a reader can legally use, in preference order, and a recommendation that resolves to a book you cannot open is not a recommendation. If the only good number is somewhere unavailable to you, the correct outcome is no bet, not a worse version of the same bet.

Worth knowing before you start: the gap between books on the same prop is not small. On one night of our board, DraftKings was the better price than FanDuel on 69.2% of the 1,350 props both quoted, and the typical row differed by 1.27 points of implied probability. More on that in do sportsbooks have different odds on player props.

What this page does not claim

It does not claim this process wins. It is a process. Any claim about results needs a sample attached, and our own pick record is being restarted as it goes public, so there is nothing here to cite and we would rather say so than point at an old number.

It is not a betting system. There is no leg construction here, no play-to price, no staking rule. The board is a form and model selector, not an engine that tells you what to bet.

What it is: the order that stops you doing twenty minutes of work on a prop the price had already ruled out.

The version with the work already done

Every step above is what the Lab board runs before it renders: the fair, the bettable price against it, the form number with its window and sample, and the matchup inputs, on every prop we track.

The point of writing out the sequence is that you can run it without us, and check ours against it.

Weighting, minimum sample and book classification are read from our own board code. Sample depth measured 12 September 2026. No sportsbook was contacted and there are no affiliate links on this page.