Points per game is the most quoted number in basketball and close to the least useful for betting. The stats that predict NBA outcomes are the ones that strip out pace and possessions, and they are all public — the edge is in using them correctly, not in finding them. For how these feed into actual markets and prices, the NBA betting guide covers the groundwork.
Why do per-possession numbers beat per-game numbers?
Two teams can both score 115 a night and be nothing alike. One plays fast and scores 115 on 105 possessions; the other grinds and gets there on 96. That is roughly 110 points per 100 possessions for the fast team against 120 per 100 for the grinder, a ten-point gap that separates a middling offence from an elite one, and per-game scoring hides every bit of it. The second team's offence is clearly better — it produces more per possession — but per-game stats cannot see the difference, because pace pollutes everything measured per game.
Offensive rating (points scored per 100 possessions) and defensive rating (points allowed per 100) fix this, and their difference — net rating — predicts future wins better than a team's actual record does. Records carry the residue of close-game luck; net rating mostly does not. When a team's record and its net rating disagree, trust the rating. The market partly prices this, but public money still bets the standings.
Pace itself is neither good nor bad. It is a volume dial, and it belongs in a different part of your process: totals.
Which stats actually predict outcomes?
Decades of analysis keep arriving at the same short list, usually called the four factors:
| Stat | What it measures | Predictive weight |
|---|---|---|
| Effective FG% | Shooting, with threes counted at 1.5x | Largest by far |
| Turnover rate | Possessions thrown away per 100 | Second |
| Offensive rebound rate | Extra shots created | Smaller than it was |
| Free throw rate | Trips to the line per shot | Smallest |
Effective field goal percentage dominates because it fixes the flaw in raw FG%: a made three is worth half again as much as a made two, and any shooting stat that ignores that is measuring the wrong thing.
Turnover rate lands second because a live-ball giveaway does double damage: it ends your trip with no shot up, and it usually hands the other team a transition bucket at the other end, so the real cost is worth more than the single possession it looks like. Free throw rate sits last, but it is the one factor a team can still lean on when its jump shot deserts it.
The rebounding note matters for anyone raised on 'rebounds win games'. As offences spread the floor and teams retreat into transition defence rather than crash the offensive glass, rebounding explains less of the outcome than it once did. It still counts. It just is not the pillar the older commentary insists it is.
How much damage does three-point variance do?
Three-point shooting is the single biggest noise source in modern basketball results. The arithmetic is blunt: a team taking 40 threes at a 36% clip will swing by about three makes — nine points — in a normal night-to-night deviation. Nine points is bigger than most NBA spreads.
Defensive three-point percentage is even noisier: how often opponents make threes against you is barely a skill at all over short samples. A defence 'allowing' 40% from deep across ten games is far more likely unlucky than broken.
For bettors this cuts two ways. Recent results are less informative than they feel — a 4-1 week built on hot shooting is mostly weather. And regression is a genuine tool: teams riding unsustainable shooting, for or against, get mispriced by a market that partly reprices results. The same variance is why live totals lag the real picture mid-game, and why losing streaks feel personal when they are not.
How do you turn these numbers into a bet?
A workable process is short:
- Start from net ratings, adjusted for schedule and rest — the back-to-back guide covers what the market already prices there.
- Use pace only for totals: expected possessions times combined efficiency is the skeleton of every total.
- Adjust for minutes and availability, which is where lineups and player-level analysis come in.
- Compare your number to the line, and pass when they agree.
Per-possession numbers will not hand you winners, but they stop you donating money to pace illusions and hot-shooting mirages, which is most of the battle. For how the markets themselves work — spreads, totals, vig — the NBA betting guide is the companion piece.