Player Efficiency Rating: What It Tells a Bettor and What It Hides

PER, in plain English

The single most common mistake I see in NBA betting forums is treating PER as a verdict. Someone posts a screenshot of a player at 24.5 PER and the replies treat it like a closing argument. It is not. PER is one number that tries to compress every box-score event a player generates per minute into a single per-game-paced rating, then standardise it so the league average always lands on 15.0. That is genuinely useful for some bet types and genuinely misleading for others, and the gap between the two is wider than most punters appreciate.

I lean on PER for two specific reads — quick player-to-player comparisons inside the same role, and futures markets where I need a baseline before going deeper. I do not use it for game-level prop research or for spread building. The reason for both decisions is the same: PER is a brilliant per-minute summariser of offensive box-score production and a poor reader of defence, context and shot quality. If you treat it for what it is, it pulls real weight in a research workflow. If you treat it as a single-number answer, it will bury you.

The Hollinger formula and its components

John Hollinger built PER in the late 1990s while writing for Sports Illustrated and ESPN. The formula has had revisions but the bones are the same. PER takes every box-score positive a player produces — field goals made, free throws made, assists, rebounds, steals, blocks — and assigns each a value in terms of expected points produced. It subtracts negatives at their points-cost: missed shots, missed free throws, turnovers and personal fouls. The result is a per-minute number, then it is adjusted for team pace and standardised so the league average comes out to 15.0 every season.

The standardisation step is what most people miss. PER is not a raw number that gets bigger when scoring goes up across the league. The formula explicitly resets the league average to 15.0 each year, which is why you cannot meaningfully compare a 2002 PER to a 2026 PER. Within a season, though, the comparison is clean. A 22.0 PER in 2025-26 sits at the same percentile as a 22.0 PER in 2010-11, even though scoring is up dramatically. That is a property worth knowing when you are looking at MVP futures or historical context — and a property worth ignoring when you are projecting a points prop on a specific night.

What sits inside the formula matters even more than the maths. The big positives are shooting efficiency, free-throw drawing, assist creation and rebounding. The big negatives are missed shots and turnovers. Notice what is not in there with much weight: shot location, shot difficulty, opponent strength, defensive positioning. The formula treats a contested heave the same as a wide-open corner three if both result in a make. That is the seed of its blind spots.

Why 15.0 is the league average

This is the design choice that gives PER its identity. By forcing the league average to 15.0 every season, Hollinger created an instant readability test. Anything above 15 is above-average. Anything in the 18–20 range is solid starter territory. The 20–22 band is All-Star quality on most nights. 25 and above is elite, MVP-vote territory. The top of the league, when one or two players are having a historic year, can push into the 30s.

The benchmark works because it gives a UK punter who is not in front of NBA games every night an instant sense of where a player sits relative to the league. If you are pricing an MVP future and you see two candidates at 27.5 and 28.1 with similar narratives, PER tells you their box-score footprints are essentially identical and you should look elsewhere — usage, team success, late-season schedule — for the tiebreaker. If you see a 24.0 and a 31.0, PER is screaming at you that the second player’s production gap is real and material. That is genuine information.

It is also worth knowing where PER stops being useful at the baseline level. Below about 13.0 PER, you are in bench-rotation territory, and the box-score signal gets noisy because minutes are small and games played are inconsistent. The numbers move around too much to anchor a bet. The reliable band for PER-anchored decisions is roughly 14 to 30, which covers most starters and stars.

When PER does help bet selection

I keep PER alive in my workflow for three specific use cases. First, season-long futures. MVP, Most Improved, scoring titles — anywhere the bet pays on cumulative box-score dominance, PER is a fair summariser of who is doing it and at what level. A futures market where Player A is 21.0 PER and Player B is 26.5 PER has probably mispriced one of them if the odds suggest a coin flip.

Second, comparing players in the same role on different teams. A starting wing on Team X versus a starting wing on Team Y, similar usage, similar minutes — PER gives a fast comparison without me having to dig through five different efficiency stats. The five player shot signatures researchers have identified from analysing nearly 60,000 attempts — Three-and-Rim, Mid-Range Master, Paint Punisher, Spot-Up Specialist and Volume Slasher — describe how a player produces. PER summarises how much he produces inside that signature. The two together give a workable read for futures comparisons.

Third, sanity-checking a hot take. If a punter on a forum or a UK commentator declares that some bench player is the most underrated guy in the league, I check his PER first. If it is 11.0, I do not need to read the next 300 words of analysis. If it is 18.5 on limited minutes, the take is at least worth a deeper look. PER is the quickest filter in the toolkit.

Defence, context and the four blind spots

This is where the article earns its keep. PER has four blind spots that punters trip over again and again, and each one matters for a different bet type.

The first blind spot is defence. PER captures steals and blocks but not the rest of defensive value — staying with a cutter, contesting a shot, executing a switch, drawing a charge that does not show up in the box score. A wing who is a genuine defensive stopper but a modest offensive player will rate 14.0 in PER. His team will outscore opponents by 10 points per 100 possessions when he is on the floor. The number says “average”. The reality says “irreplaceable”. For team-level bets — spread, totals, futures — this gap matters. The newer tracking stat work the league shipped under the NBA Inside the Game platform, including the Gravity metric that quantifies how much defensive attention an attacking player draws off-ball, is partly an answer to exactly this kind of gap. PER cannot see it. Box Plus/Minus handles defence considerably better, which is why I lean on it for game-level work.

The second blind spot is shot context. PER does not know whether a three was open or contested. It does not know whether a layup was after one dribble or seven. It does not know whether a pull-up jumper came from a broken play. Two players can put up identical PER values via completely different shot diets, and one of those diets may be far more sustainable than the other. For prop research, that matters. The high-PER guy living on heavily contested mid-range jumpers is a regression candidate. PER will not flag him.

The third blind spot is opponent strength. PER is not opponent-adjusted. A player can rack up production against weak defences and pad his rating. Across a full season this evens out, mostly. Across the small sample of a single matchup — which is exactly what a prop bet is — it does not. Reading PER and ignoring tonight’s opposition is a fast way to lose money.

The fourth blind spot is teammate dependence. PER is per-minute and pace-adjusted but it is not lineup-aware. A player whose production survives only because he plays beside an elite passer will rate higher than his standalone skill warrants. When that passer sits, his PER collapses. Pre-game, you need to know whether the rating you are reading was earned with the supporting cast on the floor or without it.

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