A friend of mine — an extremely intelligent person who watches more basketball than most beat writers — once told a room full of strangers, with complete confidence, that a player’s true shooting percentage was “basically the same thing as field goal percentage, just adjusted.” The room nodded. The room did not know what true shooting percentage was either. The argument continued for another twenty minutes, growing more heated, until somebody finally pulled out a phone and looked it up.
This is the way most sports analytics vocabulary works in practice. The terms get used. The definitions stay fuzzy. People absorb the shape of an argument well enough to redeploy it later, without ever quite knowing what they are saying. The vocabulary becomes social capital. The actual content stays optional.
The piece below is the working glossary we use at SportsHighLight before writing any sentence that contains an acronym. It is not exhaustive. It is the shortlist of terms a smart fan, broadcaster, or new analyst is most likely to hear, misread, or pretend to understand. Each entry includes what the term actually measures, where it breaks, and where to read more.
Quick read: what this glossary covers
- The headline efficiency stats — true shooting, eFG%, xG, EPA. Used in every modern broadcast.
- The methodology terms — per-possession, stabilization, regression to the mean. Where most arguments quietly go wrong.
- The intimidating acronyms — DVOA, BPM, RAPM, RAPTOR, LEBRON. What they share and where they differ.
- The terms people misuse most — “efficient” vs “productive”, “possession” vs “control”, “hot” vs “improved”.
- Where to read more — the public-facing sources that actually pay their bills.
The headline efficiency stats
These are the four metrics most likely to appear on a graphic during a Sunday broadcast. They are not perfect. They are, individually, more useful than the box score that preceded them.
True shooting percentage (TS%) is the single most useful basketball efficiency number. It folds in three-pointers and free throws and asks, in plain terms, how many points a player scored per scoring opportunity. League average sits around .576. Elite scorers live above .620. A player at .540 with high volume is hurting his team’s offense more than the box score suggests. The official definition lives in the NBA Stats glossary.
Effective field goal percentage (eFG%) is the simpler cousin. Same idea, minus the free throws. Useful when you want to isolate field shooting from foul-drawing. The two metrics travel together — read both before quoting either.
Expected goals (xG) is the soccer equivalent of an efficiency frame. Every shot is assigned a probability of being scored based on distance, angle, shot type, and defensive pressure. A team’s xG total describes the quality of chances it generated. The scoreline tells you whether it finished them. Public-data versions live at FBref and Understat.
Expected Points Added (EPA) is football’s parallel concept. Each play either improves the offense’s expected points (based on down, distance, and field position) or reduces them. EPA is the difference. It rewards leverage and punishes drive-killing mistakes more honestly than yards. The leading public dashboard is rbsdm.com.
The methodology terms that change everything
These show up less in broadcasts and more in arguments. Knowing them is the difference between citing a metric and using one.
Per-possession adjustment. Two teams scoring 110 points each in different games are not equally efficient if one played 90 possessions and the other played 105. Rate stats — points per 100 possessions, per 90 minutes in soccer, per 60 minutes in hockey — flatten out pace differences. If a writer compares offenses without naming pace, the comparison is already broken.
Stabilization. Every metric needs a sample to mean anything. True shooting stabilizes after a few hundred field-goal attempts. xG stabilizes for teams after about twenty matches. On/off splits in the NBA need thousands of possessions. A “leading the league” claim from a 12-game sample is, in most cases, noise dressed as signal.
Regression to the mean. Hot stretches and cold stretches alike tend to drift back toward an underlying skill level. A striker scoring 11 goals from 7 xG across a half-season is almost certainly going to score less in the second half. A team shooting 42% from three for two weeks will almost certainly shoot worse over the next two months. Regression is not a punishment. It is the math remembering how the world works.
Garbage time. The minutes in a game where the result is no longer in doubt. Filtering them out matters because incentives change: the trailing team starts spamming threes, the leading team empties the bench. Stats compiled across garbage-time minutes can inflate or deflate a team’s profile significantly. Cleaning the Glass built its entire NBA reputation on doing this cleanly.
The intimidating acronyms: what each one actually does
The all-in-one player evaluation metrics multiply faster than fans can keep up. The table below covers the public-facing ones a serious reader will encounter most often.
| Acronym | Full name | Sport | What it tries to measure |
|---|---|---|---|
| DVOA | Defense-adjusted Value Over Average | NFL | Opponent-adjusted play efficiency per situation |
| BPM | Box Plus/Minus | NBA | Player impact per 100 possessions, derived from box score |
| RAPM | Regularized Adjusted Plus/Minus | NBA / WNBA | Player impact controlling for teammates and opponents |
| RAPTOR | Robust Algorithm (using) Player Tracking and On/off Ratings | NBA | FiveThirtyEight’s blended model (now retired but widely cited) |
| EPM | Estimated Plus-Minus | NBA | Hybrid model combining box-score and on/off signals |
| LEBRON | Luck-adjusted player Estimate using a Box Regularized On/off model | NBA | BBall Index’s all-in-one with luck adjustments |
| PER | Player Efficiency Rating | NBA | Hollinger’s 2000s-era composite (largely deprecated) |
| WAR | Wins Above Replacement | MLB primarily | How many wins above a “freely available” player |
| xGOT | Expected Goals on Target | Soccer | xG that incorporates shot placement |
| npxG | Non-penalty xG | Soccer | xG excluding penalty shots |
Two notes on the table. First, the all-in-one NBA metrics agree on the top of the league and disagree on the borderline cases. Reading three of them and looking for consensus beats trusting any one. Second, PER is on the list because you will still hear it cited. It has been quietly demoted by the analytics community since roughly 2015 but lingers in casual coverage. Treat it as a starting point, not a verdict.
The terms people misuse most
The terms below are the ones that, in our editorial experience, cause the largest gap between what the speaker thinks they mean and what the listener hears. Each entry maps the careful version against the lazy version.
| Term | Lazy use | Careful use |
|---|---|---|
| “Efficient” | Any player scoring lots of points | High points per scoring opportunity, role and shot-difficulty adjusted |
| “Productive” | Same as efficient | High absolute output, regardless of opportunity cost or efficiency |
| “Possession” (soccer) | How much a team had the ball | How much a team controlled the territory, threat, and rest defense — different things |
| “Control” (any sport) | Dominant possession or stat-sheet edge | Whether the team imposed its preferred game state on the opponent |
| “Hot” | Player or team performing well right now | A streak that may or may not reflect actual improvement (see regression to the mean) |
| “Improved” | A team playing better than before | A team whose underlying process indicators have moved, not just its results |
| “Clutch” | A player who scores in big moments | A player whose efficiency does not drop in high-leverage situations, measured across a meaningful sample |
| “Bench mob” | A second unit that scored a lot in garbage time | A second unit whose net rating holds up against opposing rotations, not opposing reserves |
The pattern is the same in every row. The lazy version reads the surface. The careful version asks what the surface is sitting on. Most arguments about sports go better when both sides agree on the careful definition before they start arguing.
Where to read more, by sport
The public-facing infrastructure for sports analytics has matured into a small set of sites that do most of the heavy lifting. The list below is the one we would hand a new reader who wants to skip the noise.
NBA: the NBA Stats glossary for definitions, Basketball Reference for historical data, Cleaning the Glass for context-filtered analysis. Add a beat writer who watches film. That is the working stack.
Soccer: FBref and Understat for public xG and event data. StatsBomb for methodology writing. The chasm between the public-facing data and the proprietary versions is widest in soccer, but the public version is enough to argue from honestly.
NFL and college football: rbsdm.com for live EPA, Pro Football Reference for everything historical, the nflfastR package if you want to compute your own. For college, Bill Connelly’s SP+ ratings are the standard.
WNBA: public data is thinner than NBA but improving. Basketball Reference’s WNBA section and the league’s own advanced stats pages are the starting points.
For our own working introduction to how analytics fits into game-watching without becoming insufferable, the analytics hub page on this site is the entry point. The start-here page covers the editorial frame this glossary sits inside.
Frequently asked questions
What is the single most useful term to know first?
True shooting percentage if you watch NBA. Expected goals if you watch soccer. EPA if you watch football. Each is the cleanest efficiency frame in its sport and the one a serious reader is most likely to encounter inside an argument. Learn one well and the others get easier to absorb later.
Do I need to learn the math to use these terms correctly?
No. The math behind true shooting percentage is high-school arithmetic. The math behind xG is more involved but every public source publishes the outputs without expecting you to derive them. The hard part is not the formula. The hard part is the discipline to ask “compared to what?” before assigning meaning to the number.
What happens to terms when they get popular?
They tend to lose precision. PER is the textbook case — popular in the late 2000s and early 2010s, then quietly demoted as better tools (BPM, RAPM, EPM) became available. Possession percentage in soccer is in a similar position now. A term’s popularity is no guarantee of its current usefulness. Reading two or three sources that name the term’s limits is the quickest way to stay honest.
How do I tell a useful sports analytics writer from a noisy one?
By how often the writer names the limits of the metrics they cite. Useful writers say “this stat says X, but the sample is small” or “this xG number is from FBref; StatsBomb has it at a different value.” Noisy writers cite a single number as if it ended the argument. The escape hatches in the writing are the trust signal.
The takeaway, in one paragraph
Sports analytics is, in 2026, a vocabulary that has outpaced the discipline most fans bring to it. The terms are real. The misuse is widespread. Knowing what each term measures, where it breaks, and who to read for the careful version is the difference between joining the conversation and pretending to. This glossary is the working version of that knowledge. For the broader editorial frame this site uses, the start-here page is the next stop.
