Advanced Basketball Stats for Betting – Net Rating, True Shooting and the Numbers That Actually Move Lines
Available in USTable of Contents
- Why Box Scores Are Not Enough Anymore
- Net Rating – The One Number to Start With
- True Shooting Percentage – The Efficiency Number That Matters
- The Four Factors – Where Games Are Actually Decided
- Plus-Minus, RAPM and the Lineup Statistics
- Synergy and Play-Type Data
- What Advanced Stats Cannot Do
- How I Actually Use Advanced Stats Each Night
Why Box Scores Are Not Enough Anymore
I started betting basketball using the same box-score numbers everyone else used – points, rebounds, assists, shooting percentages. The results were predictably mediocre. The breakthrough in my own ROI came when I stopped looking at box scores and started looking at the advanced statistics that the sharper end of the market was already using to build their models. Once you see basketball through the lens of Net Rating, True Shooting Percentage and the Four Factors, the box score looks like a fragmentary translation of what actually happened on the floor. The advanced stats tell you the story; the box score tells you the captions.
The UK regulated market context matters here. Andrew Rhodes, then chief executive of the UK Gambling Commission, framed the scale of the domestic industry in his January 2025 speech at ICE Barcelona, noting that Gambling in Great Britain has reached the highest GGY we have ever seen – £15.6bn. Those official statistics will tell you plenty about what is going on with gambling in Great Britain.
A market that size includes a non-trivial slice of basketball volume, and the competitive pressure across UK operators means edges from advanced statistical work are smaller and more contested than they were five years ago – but they still exist for punters willing to do the analytical work.
This piece walks through the advanced statistics that move basketball lines and the ones that are interesting analytically but do not contribute to better betting outcomes. There is a real distinction between the two categories, and recognising the difference is the difference between an analyst who beats the market and a fan who quotes the latest statistical innovation while losing units.
Net Rating – The One Number to Start With
If you can only use one advanced stat for basketball betting, use Net Rating. Net Rating measures a team’s point differential per 100 possessions, which strips out the variance introduced by pace differences. A team with a Net Rating of +5 outscores its opponents by 5 points per 100 possessions, regardless of whether those possessions are crammed into a fast game or spread across a slow one. That pace-adjustment is what makes Net Rating the foundational metric – it allows direct comparison between fast-paced and slow-paced teams without distortion.
The simple version of Net Rating analysis: subtract the visiting team’s Net Rating from the home team’s Net Rating, add an adjustment for home-court advantage (typically 2 to 3 points per 100 possessions), and you have a rough estimate of the game’s expected margin per 100 possessions. Multiply that by the projected game pace divided by 100, and you have a rough margin projection for the actual game. This calculation is the kind of starting point that takes thirty seconds and gives you a number to compare against the market spread.
Where the market is generally sharp is in using full-season Net Rating as the baseline input. Where the market is less precise is in adjusting Net Rating for specific contexts – schedule density, roster availability, recent form. A team’s last-10-game Net Rating is often a meaningfully better predictor than the season-long Net Rating, particularly mid-season when rotations have shifted. The recent-form adjustment is where my own analytical work has found consistent edges, because the market tends to weight season-long numbers more heavily than the underlying predictiveness justifies.
The decomposition of Net Rating into Offensive Rating and Defensive Rating is also worth doing. Two teams with identical Net Ratings can have very different offensive and defensive profiles, and the matchup between an elite offence and an elite defence often produces a different game pattern than two average teams playing each other. The market generally handles this well at the headline level, but the secondary effects – pace, total, and prop implications of specific offensive-defensive matchups – are sometimes mispriced.
True Shooting Percentage – The Efficiency Number That Matters
True Shooting Percentage is the efficiency stat that captures all forms of scoring – twos, threes, and free throws – in a single weighted number. A True Shooting Percentage of 60 percent is elite scoring efficiency; 55 percent is good; 50 percent is league-average; below 45 percent is poor. The formula weights three-pointers and free throws appropriately, which means it does not penalise players who shoot a high volume of threes (whose conventional field goal percentage looks low) or reward players who shoot a high volume of mid-range twos (whose field goal percentage looks high).
The reason True Shooting Percentage matters for betting is that it scales cleanly with usage. A player taking 25 percent of his team’s shots at 60 percent True Shooting is contributing more to his team than a player taking 20 percent at the same efficiency, because the high-efficiency shots replace lower-efficiency shots from teammates. This trade-off is one of the inputs to expected scoring lines, and the market handles it well at the headline level but sometimes loses the nuance for individual matchups.
The specific edge I find with True Shooting is in identifying scorers whose efficiency drops meaningfully against specific defensive styles. A volume scorer who relies on rim attacks has a much lower True Shooting Percentage against teams with elite rim protection; a mid-range scorer is less affected. The market’s matchup adjustments are typically based on overall defensive rating rather than the specific defensive style, which creates opportunities on the under side of points props when the matchup style is unfavourable to the scorer’s specific approach.
The Four Factors – Where Games Are Actually Decided
The Four Factors framework, developed by Dean Oliver in the early 2000s, identifies the four statistical categories that determine basketball game outcomes: effective field goal percentage (shooting efficiency including the bonus for three-pointers), turnover percentage (turnovers per possession), offensive rebound percentage (offensive rebounds per available rebound), and free throw rate (free throw attempts relative to field goal attempts). Together, these four factors explain the vast majority of point differential variance in any given game.
For betting purposes, the Four Factors are most useful as a diagnostic tool. When a team is overperforming or underperforming its Net Rating, the Four Factors usually identify which specific factor is driving the deviation. A team whose Net Rating has improved but whose effective field goal percentage has not changed is likely getting an unsustainable bump from turnover differential or offensive rebounding – both of which are more volatile than shooting efficiency. The market sometimes treats these unsustainable bumps as predictive, which creates fade opportunities.
The framework also helps with matchup analysis. A team with elite offensive rebounding against a team with poor defensive rebounding is a matchup where one of the Four Factors meaningfully favours the offensive side, and the impact on totals and team points props is usually larger than the market accounts for. The pace and rebound interplay deserves its own dedicated workflow – the full pace-betting analysis framework covers how the rebound-and-pace combination affects total possessions and total expected scoring, which is the operational pathway from the Four Factors framework into specific betting markets.
Plus-Minus, RAPM and the Lineup Statistics
The next layer of advanced stats includes Plus-Minus, Net Plus-Minus, On-Off splits, and regression-based metrics like RAPM (Regularised Adjusted Plus-Minus) and EPM (Estimated Plus-Minus). These statistics attempt to isolate individual player impact from the surrounding context, which is genuinely difficult because basketball is a five-on-five sport where everything happens in interaction.
For betting purposes, the lineup statistics are most useful when a team has known roster changes that will alter the lineup mix in the upcoming game. If a high-impact lineup is broken up by an injury, the team’s expected performance drops by more than the simple per-game contribution of the missing player would suggest. The market often anchors on the season-long starting five even when the actual starting lineup will be different, and the predicted game outcome based on the alternative lineup can diverge from the market expectation by several points.
Synergy and Play-Type Data
Synergy Sports tracks every play in NBA games by play type – pick-and-roll ball-handler, pick-and-roll roll man, isolation, post-up, spot-up, transition, off-screen, and so on – and reports efficiency numbers for each player and team across each play type. This granular data is where the most specific matchup edges hide.
For betting, the most actionable play-type insight is identifying teams whose offensive style relies heavily on a specific play type that the opponent defends poorly. The market generally does not price play-type matchups with the same precision as headline numbers. The challenge is that Synergy data is not freely available to retail punters, and manually estimating play-type matchups from broadcast footage takes meaningful time – so the analytical investment is justified mostly for higher-volume bettors.
What Advanced Stats Cannot Do
The honest counterpart to the analytical enthusiasm above is that advanced stats have limits, and recognising those limits is part of using the stats well. Sample size is the first limitation – advanced stats become meaningful at the team level after roughly 15 to 20 games of the season, and at the individual level after meaningfully more. Early-season advanced stat analysis is dangerous because the numbers have not yet stabilised, and acting on them as if they were predictive of season-long output produces consistent losses.
The second limitation is context sensitivity. Advanced stats are calculated against a specific set of opponents and a specific roster context. When the opponent strength or roster context changes, the stat needs to be adjusted, and the adjustment is rarely as clean as the underlying number suggests. A team’s Net Rating against the bottom third of the league is different from its Net Rating against the top third, and using the average can mask meaningful predictive differences in specific matchups.
The third limitation is that the market also has access to advanced stats. The sharpest pricing on NBA games is built from models that use exactly the metrics described above, which means the easy edges from basic advanced-stat analysis have largely been arbitraged away. The current edge frontier is in the contextual adjustments – recent form, lineup changes, play-type matchups, schedule density – that the standard models do not always handle precisely. Live and in-play wagers represented approximately 47 percent of global online sports betting volume in 2024, and the live market specifically rewards punters who can integrate real-time advanced-stat reads with the in-game state faster than the live algorithm can.
How I Actually Use Advanced Stats Each Night
The workflow is fifteen minutes per slate. Step one: pull the Net Rating, Offensive Rating, Defensive Rating and last-10-game equivalents for every team playing that night, then compute the implied margin and total based on Net Rating differential and combined offensive rating. Step two: compare the implied numbers to the market lines and flag any games where the divergence is meaningful. Step three: overlay the Four Factors decomposition to understand which factor is driving the divergence and whether that factor is sustainable. Step four: check roster news and lineup adjustments that might shift the analysis.
When the workflow flags a game, I sit with it for a moment to confirm the analytical edge is real rather than a model artifact. The flags that survive that second look are the bets I take. Most nights, fewer than half my flagged games become bets, because the second look reveals injury context, line movement, or matchup adjustments that neutralise the apparent edge. The discipline to discard about half of the seemingly positive analytical reads is what keeps the actual hit rate high enough to overcome the cumulative vig.
Which advanced basketball stat is most useful for casual bettors?
Net Rating, by a significant margin. It pace-adjusts team performance so direct comparisons between fast and slow teams are valid, and the basic margin projection workflow takes about thirty seconds. The further-along stats – Four Factors decomposition, lineup data, play-type analysis – add precision but require meaningfully more analytical investment for proportionally smaller marginal gains.
How early in the season do advanced stats become reliable?
At the team level, around game 15 to 20 of the regular season. Before that, the sample size is too small for the numbers to have stabilised, and acting on them as if predictive produces consistent losses. At the individual player level, reliable signal requires meaningfully more sample – closer to 30 games for some metrics – because per-player variance is wider than per-team variance.
Are advanced stats more valuable for spread betting or totals betting?
Totals, on balance. The pace input that advanced stats provide is the single largest factor in totals projection, and pace is the variable where the market’s calibration is most often imprecise. Spreads benefit from the Net Rating framework but the sharp market is already using these inputs, while totals projections have more residual mispricing for analytical punters to exploit.
Published by the Basketball Betting Explained team.
