You are researching a number, not a player.
A sportsbook is not pricing a player. It is pricing one specific number attached to that player, and those are different questions with different answers. Research that ends in an impression of how somebody is playing cannot be compared to anything, while research that ends in a probability can be set against the price directly, and that comparison is the whole decision. This page walks the six steps in the order they actually work, from converting the posted prices to logging the wager. It also covers the two ways a careful reader still gets this wrong.
Research ends in a probability
To research a prop is to produce your own probability for a stated number and then hold it against what the market charges for that same outcome. The output is a figure such as "the over is a 55 percent shot", which is a claim that can be checked, priced, and later found to be wrong. Anything that stops short of a figure is an opinion about a player, and a player is not what is being sold to you.
That distinction is the whole reason the procedure below has an order to it. Receiving yards in football, rebounds in basketball, and strikeouts in baseball look like three different subjects, but the work is identical in all three: establish how many chances the player gets, establish what he does per chance, and convert the two into the odds of clearing one specific line. What changes between the sports is which inputs move the fastest, not the method.
If the market itself is new to you, the structure and the economics of these bets are covered in what a prop bet is. This page assumes you already know what you are looking at and covers what to do about it.
The procedure, from the line to the ticket
The order matters more than any individual step. Every one of these can be done well and still produce a bad bet if it is done in the wrong sequence, because reading a hit-rate record before you have written down an expectation is how an expectation gets written for you.
1. Start from the posted number, not from the player
Open the market before you open anything else. Read the number, read both prices, and convert them into probabilities, because the number is the thing being priced and the player is only the reason it sits where it does. Until both sides are converted you have no idea what the market is asking you to believe.
Take a receiving yards prop posted at 62.5, with the over at -110 and the under at -125. The over implies 110 divided by 210, which is 52.38 percent, and the under implies 125 divided by 225, which is 55.56 percent. Those two sum to 107.94 percent, so the book is holding 7.94 points of overround divided by 107.94, or about 7.4 percent of everything staked on the market. Rescale the pair so they total 100 and the market's fair view is 48.53 percent on the over against 51.47 percent on the under.
2. Rebuild the baseline in the right unit
A per-game average is two things multiplied together: a rate and an amount of opportunity. Only the rate is reasonably stable, so split the two apart, work in whatever unit actually produces the statistic, and multiply back up by the opportunity you expect this time. The rate is the thing you have an opinion about, and the opportunity is the thing that moves.
A receiver averaging 54 receiving yards on six targets a game is producing 9.0 yards per target. Give him eight targets because the man ahead of him is out, hold the rate exactly where it was, and the same player projects to 72 yards. That is an 18 yard move produced entirely by opportunity, with nothing about how he plays having changed. The per-game average never said which of those two numbers it was describing, and the posted line is priced on one of them.
3. Check the role, not the box score
Ask what has changed in how the player is used, because usage moves a prop faster than form does. Snap share, route participation, the spot in the batting order, and the size of a rotation are the inputs that a per-game history silently assumes have stayed the same, and they are the first things to break.
A basketball rotation that shortens from ten players to eight hands the remaining starters minutes that no per-game average from the earlier games contains, and it does so without any announcement. Baseball changes the same quantity more visibly, because a hitter who moves up the order takes an extra turn at the plate in a meaningful share of games, and four plate appearances becoming five is 25 percent more opportunity before anything about the hitter is in question.
4. Adjust for the opponent and the conditions
Only now does the matchup enter. Pace decides how many possessions exist, the expected game script decides how those possessions get spent, and in outdoor sports the weather decides how far the ball travels. Each of those changes the opportunity, the rate, or both, and they do not always push in the same direction.
Game script is the input people skip. A team favored by two touchdowns spends the fourth quarter running the ball and protecting a lead, so the same forecast pushes a rushing prop up and a receiving prop on the same roster down. Basketball has the mirror image in blowout risk, where a starter's minutes get cut precisely because the projection was right. Outdoors, wind at an open ballpark and heavy weather in football change how far the ball travels, and there is more on that in the guide to totals.
5. Turn the read into a probability and beat the price
State a number you can be wrong about. A lean toward the over is not a number, and only an actual probability can be set against a price and shown to be worse than it. Then check it against two benchmarks: the fair de-vigged number, and the break-even rate of the price you would really be taking.
Return to the 62.5 yard prop. Suppose your work says the over is 51 percent. That beats the market's fair number of 48.53 percent by nearly two and a half points, which feels like an edge, and it is not one. The price you would take is -110, whose break-even is 52.38 percent, so the wager returns 0.51 multiplied by the decimal price of 1.909, which is 0.974 per unit risked and a loss of about 2.6 percent. At 55 percent the same wager returns 1.050, a gain of 5.0 percent. The 3.85 points between the fair number and the break-even rate is the toll, and disagreeing with the market by less than the toll is not a bet.
6. Shop the number and both prices, then size it and log it
Prop markets disagree across books far more than main markets do, and the disagreement is often in the number itself rather than only in the price. Take the best available version of the wager you have already decided to make, stake it the same as every other prop you bet, and write down the price you got.
Finding the under at -115 instead of -125 drops the break-even rate from 55.56 percent to 53.49 percent, a gain of 2.07 percentage points on a wager you had settled on. Props are not special here, and the honest comparison makes that clear: ten cents shopped on a main side, from -115 to -105, is worth slightly more at 2.27 points. What makes it pay here is that a ten cent gap is routinely sitting there on a prop, where main sides at the large books converge to within a cent or two, which is why shopping the price pays better on props than anywhere else. Watch for the number too, since one book at 62.5 and another at 64.5 is a far larger difference than any price.
Being right about the number is not enough
Most prop research fails at the last step rather than the first, and it fails by measuring itself against the wrong benchmark. Beating the market's fair number feels like being right, and it is not the same thing as being paid. On the 62.5 yard prop above, the fair view was 48.53 percent and the price being taken demanded 52.38 percent, so everything between those two figures is a toll rather than an edge.
That toll is what makes props punishing, because it is far larger here than on the markets the same bettor is used to. A standard side at -110 on both sides has a fair view of 50 percent against a break-even of 52.38 percent, so the toll is 2.38 points. On this prop it is 3.85 points, roughly sixty percent wider, for the ordinary reason that the book holds 7.4 percent of the market instead of 4.55 percent. Research of exactly the same quality therefore clears the bar on a side and fails on a prop, which is why prop work has to be better, not merely equal, to be worth the same money.
The practical version is a single question asked in the right order: not whether the market is wrong, but whether it is wrong by more than the toll. Anything short of that is a correct opinion and a losing ticket. Recent hit rates are the other trap here and what a prop bet is takes them apart properly, while the prop trends board plots each result against the line that was actually posted at the time.
The mechanism test
Write your own number down before you open the history, for the anchoring reason that what a prop bet is sets out in full. What that page does not give you is the rule for deciding which of the props that survive those six steps, the ones whose own probability clears the break-even, are actually worth a stake. It is a single question rather than a checklist.
Then apply one fixed pass rule to everything that survives: if you cannot name the mechanism in a single sentence, there is no bet. "The starting middle infielder is out and this hitter moves up two spots in the order" is a mechanism, because it names a cause and predicts a direction. "He has gone over in four straight" is not a mechanism, it is a description of the past with no claim inside it. Roughly speaking, a mechanism survives being told to somebody who disagrees with you, and a trend does not.
The rule also does useful work in the other direction. When the mechanism is nameable but the price has already absorbed it, you get to pass quickly and move on, which is most of what a research process is for. Turning the read into a probability and checking it with the no-vig calculator takes under a minute, and implied probability covers the conversion itself in detail.
What gets shortened is the method, not the count
This procedure has a failure mode of its own, and it is not laziness. An hour of genuine work on four props produces a feeling of authority that does not stay attached to those four props. The bettor who has earned a view on four ends up with positions on forty, and the forty were not researched, they were adjacent to research. What was shortened was never the number of tickets, it was steps two through five.
The defence is to fix the count before the slate opens rather than while it is running, and to treat it as the harder of the two limits to hold. A stake size announces itself when it is breached, because the figure on the slip is visibly larger. A count does not, because every individual addition is small and each one arrives with a reason attached. If the number of props you are betting has drifted well past the number you actually worked through, the responsible gambling page collects the limit and self-exclusion tools in one place.
Researching props, answered
How do you research a player prop?
Start with the posted number and both prices, convert them into probabilities, and only then work out what the player's underlying rate should be. Build that rate per opportunity rather than per game, adjust it for the role the player is in now and for the opponent and the conditions, and turn it into a probability for the exact number in front of you. The research is finished when you hold a number that can be compared to a price, and not before.
Are last five game hit rates useful for props?
A short hit rate records what a player did against whatever numbers were posted at the time, which is rarely the number posted now. If the line has moved since, that record is answering a question nobody is asking. The individual game results are still useful as raw counts you can re-score against the current number yourself, but the packaged record on its own is close to worthless.
Which statistics matter most for player props?
The statistics that measure opportunity matter most, because opportunity moves a prop further than efficiency does. Targets and snap share in football, minutes and usage in basketball, and plate appearances and the batting order slot in baseball are what to check first. Efficiency numbers matter after that, and over a handful of games they are far noisier than most bettors treat them as being.
How can you tell whether a prop line is wrong?
You cannot tell whether a prop line is wrong by looking at it. You can only tell by having your own number to hold against it, which means converting both sides of the market, removing the margin so the two probabilities sum to 100 percent, and putting your estimate beside the result. A line is worth attacking when your estimate clears the break-even rate of the price you would be taking, which is a higher bar than simply disagreeing with the fair number.
Should you bet a lot of player props?
Volume is the most reliable way to turn a small prop edge into a loss. Honest research on one prop takes real time, a normal slate offers hundreds of them, and the number a person can genuinely work through is much smaller than the number available. Betting the ones you researched and passing the rest is a method, while betting the rest because they are on the screen is not.
What is the single biggest mistake in prop betting?
The biggest mistake in prop betting is doing all of the work on the player and none of it on the price. A bettor can be right about which way a number should move and still lose across a season, because the estimate has to clear the break-even rate of the price actually being taken, not merely disagree with the market's fair number. On a prop holding 7.4 percent those two benchmarks sit 3.85 points apart, and research that lands in the gap between them produces a confident losing bet.
How does an injury elsewhere in the lineup change a player prop?
An absence redistributes opportunity, and opportunity is most of what a prop is made of. A missing starter frees up targets, minutes, or a place in the batting order, and the players who absorb them see their expected production change before anything about their own form does. The prop on the absent player moves fast, because one player is the whole market and there is nothing else in it to absorb the news. What moves slowly is the redistribution, since the teammates who inherit that opportunity are priced on histories compiled without it.
See the line and the result on the same chart
A hit rate quoted on its own hides the number that produced it. Our prop trends board plots a player's line, the market's prices, and the actual results together, so step one of this procedure takes seconds instead of tabs.