NBA Player Props Research: A Practitioner's Framework for Finding Real Edges

Why prop research pays better than picking favourites

Three winters ago I sat with a spreadsheet of 84 points-props from a single Wednesday slate and worked out, retroactively, that 11 of them had been priced more than two points off the median I would have projected. That is a ridiculous number. In side-and-total markets, the equivalent slate offers maybe one mispriced line if the gods are smiling. Props are different. The books take more shots, hedge less aggressively per market, and update slower when news lands. For a careful researcher, that translates into edges that simply do not exist in the sides game.

The reason is structural. A book has to balance one side market against the other across a fixed pool of money. A prop is a thin, single-sided market with limited liquidity and a much smaller pool of sharp money correcting it. Even sophisticated modelling teams cannot model every player on a ten-team night with the same depth they model a single spread. So the market has gaps. Your job, as a punter, is to find them.

This article is about how. Not which props to bet tonight, but the framework I use every night to find them. It starts with two inputs the book cares about more than the public realises — usage rate and projected minutes — and ends with a research routine that survives contact with the actual NBA calendar. We will also spend time on the Jontay Porter saga, because every prop bettor in 2026 needs to understand what changed when the league handed down its first lifetime gambling ban in modern history.

One framing line before we start. A points prop is not a guess about a player’s ceiling. It is a market estimate of his median scoring output given expected minutes and expected role. To beat it, you do not need to predict greatness. You need to project the median more accurately than the book does. That is a much smaller, much more winnable task.

Usage rate as the first thing you check

Of all the inputs I look at on a prop, usage rate is the one I check first. Always first. The reason is brutally simple: a player who does not touch the ball cannot score, no matter how efficient he is, no matter how many minutes he plays, no matter how favourable the matchup looks on paper.

Usage rate measures the percentage of team possessions a player ends while on the floor — through a field-goal attempt, a free-throw trip or a turnover. The league baseline is 20 per cent, since five players share 100 per cent. A featured guard or wing runs at 28 to 32 per cent. A primary star, the kind whose name moves prop markets on its own, runs at 32 to 36 per cent on starter minutes. Anything above 36 is exceptional and unstable. Anything below 16 is a 3-and-D role player who scores in spurts.

The reason usage anchors the projection is that it links directly to possessions. With league pace sitting around 104.5 possessions per 48 minutes in 2025-26 — a 30-year high — a player on the floor for 36 minutes is being exposed to roughly 79 team possessions. At 28 per cent usage, that is 22 ended possessions in his pocket. Multiply by a points-per-possession rate derived from his true shooting and you have a projection.

The mistake new prop bettors make is using last season’s usage. Usage is the input most affected by lineup changes and by what minute group a player runs with. A wing who played 30 per cent usage with his second unit last season might drop to 22 per cent when he plays starter minutes against the opposing starters. The same wing, when his team’s star is rested, can spike to 34 per cent for a single night. The trailing five-game usage, broken down by minute group, is the version that predicts. Season-long usage is the version that misleads.

The deeper reason usage matters is that books increasingly price it correctly at the top of the slate. They do not price it correctly down the slate. A featured star is hard to mis-price on a points line because his usage is famous. A rotation big who jumps from 18 to 26 per cent because a teammate is out is a different story — that change is where the soft prop lines hide. For the deeper mechanics of how usage moves, including the lineup-by-lineup variance and the way usage redistributes when stars sit, I have written a focused piece on NBA usage rate that walks through the modelling in detail.

Minutes projections, the silent killer of prop models

The number that ruins more prop models than any other is not usage. It is minutes. Get the minutes wrong by four and your entire projection is broken before you do any other work.

Minutes are tricky because they are not predictable from box scores alone. They are produced by a coach’s rotation pattern, by foul trouble, by blowouts, by load management and by injury contagion. A starter who normally plays 34 minutes will play 28 if his team is up 18 entering the fourth quarter. The same starter will play 39 if the game is tied with five minutes left in the third. The prop line is anchored to the expected case, which is roughly the rotation median in a competitive game. If you can identify games that are likely to deviate from that median in a predictable direction, you have an edge.

Three signals dominate. The first is coach pattern: some coaches rest their starters early when leading and others do not. Track this empirically over the last 20 games of each team and you will find consistent biases. The second is opponent quality: a team with a 14-point spread against a weak opponent is more likely to produce a blowout, which compresses starter minutes. The third is rest, and this is where prop bettors who model rest correctly get paid. The relationship is tight: every additional rest day reduces injury probability by about 16 per cent, while every 96 minutes of accumulated workload raises it by 2.87 per cent. Coaches know these numbers. So do team trainers. A heavily-worked starter on a back-to-back is far more likely to be capped at 28 or 30 minutes than the same starter on three days off.

The practical adjustment looks like this. Start from a player’s trailing 10-game minutes average, weighted toward games that match the current rest profile. Then adjust for opponent: if the matchup spread is greater than 12 points and the player is on the favoured side, knock two to four minutes off. If the player is on a back-to-back and his coach has a documented pattern of capping minutes in that situation, knock another two off. The result is a minutes projection that beats the trailing average by enough to move prop lines materially.

One last point. Foul trouble is mostly noise — you cannot predict it pre-game. But certain matchups have foul-rate fingerprints. A big guarding a foul-drawing star is more likely to pick up early fouls than the same big guarding a spot-up shooter. Marking those matchups, and noting which bigs play through fouls versus get pulled, is the kind of detail that separates an amateur prop model from a professional one.

Building matchup-adjusted projections that hold up

Here is the loop I run for every prop I price. Five inputs, in this order: minutes, usage, true shooting, opponent-allowed efficiency at the player’s position, and pace adjustment. Five numbers. Multiply them in the right way and you have a projection. Compare to the line and you have a position.

The math is simpler than it sounds. Take a wing projected for 33 minutes and 26 per cent usage against a defence that allows league-average pace of 104. That gives him roughly 18 to 19 ended possessions in his usage. At a trailing 15-game true shooting of 58 per cent, his expected points per ended possession is around 1.16. Multiply: 18.5 x 1.16 ~ 21.5 expected points. If the book has him at 19.5, you are looking at a clear over. If the book has him at 22.5, the line is sharper than your projection and you stand down.

The matchup adjustment is where the real work is. Two defences allowing the same league-average effective field goal percentage can produce wildly different prop outcomes if one funnels wings to the mid-range while the other lets them shoot threes. Most public dashboards now publish opponent-allowed eFG% and TS% by position group, and the better ones go further into shot type. A defence allowing 62 per cent eFG% to shooting guards inflates the wing’s expected TS%; a defence allowing 49 per cent deflates it. The matchup-adjusted TS% is what you actually multiply, not the player’s raw rolling number.

Shot diet matters too. The five shot signatures derived from a 59,227-shot analysis — Three-and-Rim, Mid-Range Master, Paint Punisher, Spot-Up Specialist and Volume Slasher — give you a fast read on how a player generates points. A Spot-Up Specialist needs ball movement to score; a Volume Slasher creates his own looks regardless. When a team loses its primary creator, the Spot-Up Specialist’s prop drops harder than the box score adjustment would suggest. The Volume Slasher’s prop barely moves. Knowing which archetype you are pricing is the difference between catching a soft over and getting buried by a stale projection.

Pace is the multiplier I add at the end. The league average sits at 104.5 possessions per 48 minutes, with team-to-team variation of about six possessions top to bottom. A wing playing against a top-pace opponent gets about six per cent more possessions than the same wing against a slow team. That alone moves a 21-point projection to 22.3 — enough, on its own, to push a marginal under into a clean over.

The whole loop, done properly, takes about six minutes per prop. Multiply by ten props a night and you have an hour of focused research producing a board of priced lines. Compare to the market and you act on the gap. Anything inside half a point of the book’s line, I leave alone. Anything outside a full point, I take a position. That filter alone is most of what separates a long-run winner from a long-run loser in prop markets.

Line shopping and why it matters more in props

If you only do one thing differently after reading this piece, do this: line shop every prop, every night, across at least three books. Props are the market where line shopping is least optional, because the inter-book disagreement is larger than in any other NBA market.

The reason books disagree on props is the same reason books offer props at all. Each book models a small subset of the league deeply and the rest of the league shallowly. The shape of the disagreement is unpredictable: one book may carry the soft side of a points line while a competitor has the soft side of the assist line for the same player on the same night. Without comparing both, you are accepting the first price you see, which over time is the equivalent of giving up half a point of expected value per bet.

The number that matters here is closing line value, or CLV. The strongest single predictor of long-run prop profitability is not your raw win rate. It is whether you are consistently beating the closing line at the book you actually bet. A 53 per cent win rate against the closing line is a profitable prop bettor; a 58 per cent win rate behind the closing line is a losing one. The market is the test, not your record on any given night.

The practical move. For every prop on your card, log the line you bet, the closing line, and the win-or-loss. Once you have 200 of those rows you can compute your CLV per market. If your CLV on, say, rebound props is positive and your CLV on assist props is negative, drop the assist props from your card. Track the source, not the symptom. The win rate will sort itself out over a thousand bets; the CLV tells you in 50.

The Porter case and why the integrity context matters now

On 17 April 2024 the NBA banned Toronto Raptors player Jontay Porter for life. The proximate cause was an $80,000 wager placed by an associate on a parlay that included Porter’s own under-prop performance for that game, which produced a payout that would have totalled $1.1 million if not flagged and frozen by the books before settlement. A separate account, run by Porter himself or by intermediaries acting for him, placed 13 bets sized between $15,000 and $22,000 across 2023 and 2024 and netted $21,965 in profit. That second number is the more telling one. It is small money in betting terms. It is career-ending money in NBA-employee terms.

When Adam Silver announced the ban, he framed it in the strongest language the league office had used on a gambling matter since the integrity reforms began. “The cardinal sin of professional sports is gambling on games,” Silver said in his April statement, drawing an explicit line between conduct the league would tolerate as serious and conduct it would not tolerate at all. The phrasing was deliberate. It signalled a regulatory posture that has shaped the prop-market landscape ever since.

The first downstream change was in how books and the league treat two-way and end-of-roster players in prop markets. Players on two-way contracts and short-term deals are far more exposed to the kind of financial pressure that produced the Porter case, and their prop lines are now subject to heavier review, lower limits, and in some sportsbooks outright unavailability. The second change was in book settlement protocols — the Porter parlay was caught because the unusual volume on a thin market triggered a freeze, and that integrity layer is now more aggressive across the prop landscape. You will see this most often as suspended markets when news breaks, or as lower per-bet limits on props involving low-minutes players.

For a punter, none of this changes the math of finding edges. It changes the playable surface. Markets that used to exist do not exist any more. Books that used to take five-figure prop bets now take three. The professional response is to focus on starter-and-rotation props, where the integrity layer is lighter because the markets are deeper, and to accept that the edge cases at the bottom of the roster — which were always thin anyway — are mostly gone.

The broader point, the one I want to land here, is that the prop market is a regulated market now in a way it was not five years ago. The wild west pricing of 2018 is over. The edges are smaller, the books are sharper, and the integrity infrastructure is layered. That is not a reason to stop researching props. It is a reason to research them harder, with cleaner models, on the markets where the work still pays.

A weekly research workflow that survives the season

The single biggest mistake I made in my first two years of prop work was treating every night as a fresh problem. Eighty-two games is too many for that. The season is a marathon of compounding small decisions, and the people who do well are the ones with a routine that holds up in February when motivation is lowest. Here is mine.

Sunday night, I rebuild my baselines for the coming week. That means refreshing trailing 15-game numbers for every player on the slate, refreshing opponent-allowed efficiency tables, and noting which teams have schedule pressure — back-to-backs, four-in-six stretches, long road swings. This is the slow work and it cannot be done on game day. I block 90 minutes for it and it stays in the calendar regardless of mood.

Monday through Sunday, the daily routine is shorter. About 75 minutes, broken into three blocks. Block one, an hour before lines post: read the injury reports, check the rotation patterns from the last three games, note which teams are travelling. Block two, when lines post: scan the prop board and flag any line that diverges from my Sunday baseline by more than 1.5 points in either direction. Block three: price the flagged lines properly, against the matchup-adjusted projection, and bet only the ones outside a full point of the line.

The discipline is in block three. The temptation, especially in February, is to bet the flagged lines without doing block three. Do not. The flag is the alert; the projection is the bet. Skipping the projection is how you turn a profitable system into a slow-leak one.

One more discipline. Keep a journal of bets, not just outcomes. For every bet, log the line, your projected number, the CLV at settlement, and a one-sentence reason. After 200 entries you will see patterns in your own work that you cannot see in week-by-week P&L. The journal is the closest thing to a coach a solo bettor has.

The season is long. The market is sharper than it was five years ago. The edges are smaller and the integrity layer is heavier. None of that changes the basic shape of the work, which is to project medians better than the book and bet the gap. Usage and minutes give you the volume. True shooting and matchup-adjusted efficiency give you the conversion. Pace and shot diet give you the multipliers. Line shopping captures the value. CLV measures the result. That is the framework, and it survives the calendar.

Player props questions UK punters ask

Created by the "NBA Stats For Betting" editorial team.