True Shooting and Effective Field Goal: The Two Stats That Beat Raw FG%

Why FG% lies and what to use instead

For about a year early in my analyst days, I built models around field-goal percentage. Just FG%. It is the headline number on every broadcast, and I treated it as a serious efficiency signal. I lost money for that year. Not catastrophically — I had pace right, I had rest right, I had matchups right — but consistently enough that I knew something in the engine was wrong. The thing that was wrong was using raw FG%.

FG% counts a 24-foot three the same as a tipped-in put-back. It rewards two-point grinders and punishes the modern shooting offence. It treats free throws as if they did not exist, even though a team can score 25 points a game from the line. And once you understand what FG% leaves out, you understand why every serious NBA model uses one of two alternatives: effective field-goal percentage or true shooting percentage.

These are not interchangeable. eFG% solves the three-point problem. TS% solves the three-point problem and the free-throw problem. Each has a place in betting work. Each has a specific failure mode. Once I switched from FG% to building my reads around eFG% for team-level totals and TS% for player props, the chronic small bleed in my P&L closed.

What follows is the version of this lesson I wish I had read in 2017. We will define both metrics, walk what they actually capture, separate the situations where one beats the other, and look at how they translate to the markets that UK punters bet most — totals, spreads and props. We will end on the four mistakes I see other punters make with TS% specifically, because the metric is so good at suggesting confidence that it can quietly lead you astray.

The headline: FG% is a number for highlight reels. eFG% and TS% are numbers for pricing bets.

Effective Field Goal Percentage: the three-point fix

Effective field-goal percentage was the first repair to FG%. The thinking is simple: a made three is worth 1.5 made twos, so any percentage that treats them equally is throwing away information. eFG% adjusts the math to fix that.

The formula is:

eFG% = (FGM + 0.5 x 3PM) / FGA

Field goals made, plus half a credit for each three made, divided by total field-goal attempts. The half-credit is the conversion factor that makes a three worth 1.5 twos in the numerator while keeping the denominator on a per-attempt basis. The result is a percentage that scales properly with shot value.

A worked example makes the point. Team A goes 40-for-90 from the floor with 12 threes made. FG% = 44.4 per cent. Team B goes 40-for-90 from the floor with 0 threes made. FG% also 44.4 per cent. The two teams have produced the same FG%, but Team A has scored 92 points to Team B’s 80. eFG%, computed properly, gives Team A 51.1 per cent and Team B 44.4 per cent. The seven-point gap is exactly the spread of efficiency that FG% hides.

That gap is what was costing me in 2017. I was reading two teams with identical FG% and treating them as equivalent shot producers. They were not. The team firing more threes was, in dollar terms, the better offence — even when the make rate looked the same.

In 2025-26 the league averages roughly 53 to 54 per cent eFG%, up from the high 40s a decade ago. That structural shift tracks the three-point revolution. Teams take more threes, value the catch-and-shoot ratio more highly, and design offences to maximise eFG% rather than FG%. The metric has become more useful as the league has organised around it.

Where eFG% still falls short: free throws. A team that drives, draws contact and shoots 30 free throws a game looks identical to a team that drives, draws no contact and shoots 18 free throws a game, if their field-goal selection is the same. eFG% does not see free throws. For modelling totals, this matters less than you might think — free-throw rate enters separately into most possession-level models. For modelling player props, it matters a great deal.

The clean read: eFG% is the right metric when your question is “how efficient is this team at converting its non-free-throw shots.” That covers most spread and totals work in 2025-26, where shot selection is more decisive than free-throw rate.

True Shooting Percentage: including the free-throw line

True shooting percentage takes eFG% one step further. It folds in free throws, and in doing so it becomes the single best one-number measure of scoring efficiency in basketball.

The formula:

TS% = Points / (2 x (FGA + 0.44 x FTA))

Total points scored, divided by twice the number of “true shooting attempts” — field-goal attempts plus 0.44 times free-throw attempts. The 0.44 coefficient should look familiar: it is the same one used in the pace formula, for the same reason. It approximates the share of free-throw trips that end the possession.

The metric is denominated in a way that lets you compare any scorer in any era. A player shooting 60 per cent TS is scoring at a 60 per cent clip per true shooting attempt. League average TS% in 2025-26 sits around 58 per cent. Anything north of 60 per cent on meaningful volume is elite. Anything below 55 per cent on a starter’s usage rate is a problem.

The example that made TS% click for me, years ago, was the contrast between a high-usage isolation guard and a roll-and-finish big. Same age. The guard shot 44 per cent from the floor, 35 per cent from three, 86 per cent from the line. The big shot 64 per cent from the floor, 0 per cent from three, 56 per cent from the line. eFG% gave them roughly comparable scoring efficiency. TS% told a different story. The guard, shooting nine free throws a game, ended up with 58 per cent TS. The big, shooting four free throws, ended up at 65 per cent. Per scoring opportunity, the big was the more efficient player by a wide margin. Their teams’ point-per-possession numbers reflected that.

When does TS% beat eFG%? Any time free throws are a meaningful share of the offence. Whole-game models always use TS%. Player prop models, especially for high-volume scorers who live at the line, always use TS%. The exception is when you want to isolate shot selection from foul-drawing — a specific question for which eFG% is the cleaner tool.

The big shooting analysis project that surveyed 59,227 NBA shots and identified five distinct shot signatures — Three-and-Rim, Mid-Range Master, Paint Punisher, Spot-Up Specialist, Volume Slasher — used eFG% to separate the signatures themselves but TS% to rank the players within each signature. That is the standard professional workflow: eFG% to characterise, TS% to value.

One last caveat. TS% rewards getting fouled. It does not directly reward drawing fouls without making the free throws. A 60 per cent TS player who shoots 70 per cent at the line is producing different value than a 60 per cent TS player who shoots 90 per cent. Both numbers should be in any prop model that uses TS% as a primary input.

TS% vs eFG%: when to use which

Here is the question I get most often: when do I use one and when do I use the other?

The boring answer is that any professional bettor uses both. eFG% is the cleaner read on shot-selection efficiency. TS% is the more complete read on total scoring efficiency. Which one matters depends on what you are pricing.

For game totals, use eFG% as your headline efficiency input. Totals models are forecasting team-level points, which are produced by the combination of pace and per-possession scoring. The per-possession scoring number is dominated by field-goal efficiency, which eFG% captures directly. Free-throw rate enters separately, as a possession-level adjustment, and folding it into TS% at the team level mixes two signals that the model wants kept apart.

For player props, use TS%. Props are forecasting individual scoring totals, where free-throw production is a meaningful chunk of the output and where the question is “how many points does this player score per opportunity.” TS% answers that question in one number. eFG% requires a free-throw adjustment downstream.

For spread work, eFG% differential is a stronger predictor than TS% differential, because the per-shot efficiency gap is the thing that compounds across the game. But the difference is small. If you have only one number to work with, either will do.

For evaluating individual shot diets — the question of what kind of scorer a player is, not how productive they are — eFG% wins. TS% washes out the structural choice of shooting threes versus drawing fouls, because both raise TS%. eFG% separates the two paths and tells you which one the player is actually walking.

A useful mental model: eFG% is the lens for what kind of offence is being run. TS% is the lens for how well that offence is producing points. Sometimes the answer is “an efficient three-point offence producing well.” Sometimes it is “an inefficient three-point offence producing well anyway because the foul-line rate is enormous.” The two metrics together can distinguish those cases. Either alone cannot.

Shooting efficiency and totals markets

There was a stretch in late November when I logged twelve totals that all moved against the public by two to three points before tip-off. Pace numbers were holding steady. The driver each time was efficiency. Specifically, eFG% trends over the previous week, which the sharps were pricing in and the casual money was missing.

Efficiency drives totals more than pace does at this point in the season. Here is why. League pace is now stable around 104.5 possessions. Team-to-team pace variation is roughly six possessions from top to bottom. League eFG% is stable around 53 per cent, but team-to-team eFG% variation is closer to seven percentage points from top to bottom. Convert each into points: a six-possession pace swing produces about ten points of total. A seven-point eFG% swing produces about 14 points of total at constant pace.

Put differently: efficiency variation across teams is bigger than pace variation, and books price it more aggressively than the public does. The total moves that show up overnight are usually efficiency moves, not pace moves.

The league offensive rating of 114.3 points per 100 possessions is built on that eFG%. If you see a total that implies a per-possession scoring rate dramatically out of line with the eFG% of the two teams, look harder. Either the book has information you do not — usually injury news — or the line is genuinely wrong. The first case is much more common than the second.

Where efficiency-driven totals get tricky is the Q4 shooting drop. I mentioned earlier that the Cohen’s d on shooting accuracy from Q1 to Q4 sits at -1.27, which is huge. That decay does not show up in season-long eFG%, but it shows up in actual game scoring. A team that posts a 56 per cent eFG% over 82 games is averaging across quarters. In close games, the Q4 efficiency they actually deliver is materially lower. Live totals markets respect this. Pre-game models often do not.

The practical move: if a game is likely to be close — and roughly 19 per cent of NBA games are within ten points entering Q4 — discount your team eFG% inputs by something like five to seven percentage points for the closing quarter when projecting full-game totals. It is not pretty, but it captures the decay that season-long numbers wash out. Doing this consistently moved my totals win rate noticeably; not enough to be statistically clean, but enough to keep the routine.

Reading prop lines through TS%

Prop markets are where TS% earns its keep. A player’s TS% over the trailing 15 games is the single best public-data input to a points-prop projection. Not the only input — pace and matchup matter — but the headline.

The mechanics. A prop line for points is the book’s estimate of the player’s median scoring output, with juice on either side. To beat it, you need to project the player’s actual median, then compare to the line. TS% gives you the conversion rate. Combined with usage rate and minutes — both of which I cover in the props strategy work — TS% closes the loop from “how many opportunities does this player get” to “how many points does that produce.”

A worked example, simplified. A guard projects to 35 minutes and a usage rate of 26 per cent against a defence allowing a league-average pace of 104. That gives him roughly 19 to 20 true shooting attempts in the game. If his trailing 15-game TS% is 60 per cent, the expected scoring is around 22 to 24 points. If his TS% is 53 per cent, the expected scoring is 20 to 22. That two-to-three-point swing in expected output, against a typical prop line of 18.5 or 19.5, is the difference between an over and an under play.

Where this gets sharper is matchup-adjusted TS%. The same guard, against a defence that allows 62 per cent eFG% to shooting guards, has an expected TS% above his baseline. Against a defence that allows 49 per cent, below. The matchup adjustment is where edges live. Most public dashboards now expose opponent-allowed efficiency by position; the work is to plug it in.

The thing I see new prop bettors do wrong is overweight points-per-game and ignore TS%. A player averaging 22 points might be doing it on 18 shots at 56 per cent TS, or on 24 shots at 50 per cent. Same output, very different expected variance. The 56 per cent player is more reliable. The 50 per cent player is a coin flip even when the line looks easy.

Shot diet matters too. The five shot signatures derived from the 59,227-shot study give you a quick read on how a player produces. A Spot-Up Specialist needs movement to score; a Volume Slasher does not. Their prop ceilings react differently to lineup changes. For deeper context on how shot selection moves prop variance, I expand on this in the three-point rate analysis.

Team-level vs player-level efficiency

The trap I see most often is using team eFG% to project player eFG% or vice versa. They are different objects.

Team eFG% is an emergent property of five-player rotations. It is the weighted average of every shot the team takes, across roles. A team that posts a 56 per cent eFG% might have a Three-and-Rim specialist shooting 68 per cent, a Mid-Range Master shooting 47 per cent and three Spot-Up Specialists in the middle. The team number tells you about the offence as a system. It does not tell you about any individual player.

Player eFG% is the underlying signal. It tells you what kind of shooter a player is, full stop. League average for non-centres is roughly 52 to 54 per cent. League average for centres, who get more rim attempts, can climb above 60 per cent.

The mistake is moving across the levels without translating. A team that adds a high-eFG% specialist usually does not see its team eFG% rise by the difference between the new player’s number and the average; the new player takes shots that someone else used to take, and the team-level rise depends on which shots they were. Diminishing returns are real.

For betting work, the levels separate cleanly. Team eFG% goes into totals models. Player eFG% goes into prop models. Crossing the streams is what produces the worst kind of mistake — overconfidence based on accidentally double-counting an efficiency signal across two markets.

The exception is when a team’s eFG% is anchored almost entirely by one player’s volume. A team with a 30-usage guard producing 60 per cent of the offence will move with that player’s individual eFG% in a way that, mechanically, a team with five 20-usage players will not. In those cases — typically two or three teams a year — team and player efficiency become correlated tightly enough that the streams cross. You can use either. But you have to know which case you are in before you simplify.

The four common misuses of TS%

Four traps. I have fallen into each at least once.

First, treating TS% as static. Players’ TS% drifts within a season for reasons ranging from health to shot selection to opponent-allowed efficiency. The trailing 15-game TS% is more useful for prop work than the season-to-date number, and the trailing five for live work. Anchoring on season number when the trailing window is telling a different story is a slow leak.

Second, confusing high TS% with high value. A bench player can post a stratospheric TS% because his attempts are cherry-picked: open threes off ball movement, rim runs after defensive switches. The TS% does not translate to a starter’s role, where his attempts would be contested. Volume matters. Shane Battier put it well — “Analytics is like blackjack. When the dealer has a five showing, what do you do? You double down. Why? Because the book tells you that is the best play at the time and gives you the most chance to win the hand and win money.” That logic applies to TS% with usage attached, not TS% in a vacuum.

Third, comparing TS% across positions without adjusting. Centres post higher TS% than wings on average because they get easier shots. The cross-positional read is misleading. Compare TS% within position group, or use a position-adjusted version like TS+ where available.

Fourth, ignoring the free-throw component. A player with 60 per cent TS who shoots 60 per cent at the line is producing differently than the same TS% with 90 per cent shooting. The free-throw split matters for variance, especially in live markets where the foul rate climbs in close fourth quarters.

TS% and eFG% questions punters ask

Published by the NBA Stats For Betting team.