xG FAQ
Every question answered in the guide, in one place, each linked to the chapter that explains it in full.
Chapter 1 · What is xG?
What does 1.5 xG mean?
The chances a team created were good enough for about one or two goals on an ordinary day. It is an average over many similar chances, not a forecast for this match. Read the chapter →
Is a high xG the same as a good performance?
Not on its own. xG measures chance quality. A team can create a lot and defend badly, or protect a lead and shoot little. Read it with xG against. Read the chapter →
Who invented xG?
The idea of judging shots by how often similar shots are scored goes back to academic work in the 1990s; Opta introduced its xG in 2012, and TV brought it to a wide audience from 2017. Read the chapter →
Chapter 2 · How xG is calculated
Is there a formula for xG?
Not one formula. Old models were simple regressions with a weight per input; today's are machine-learning models trained on hundreds of thousands of shots. Each provider has its own. Read the chapter →
Does xG include the goalkeeper?
In premium models, the keeper's position is an input. The keeper's skill is not: that belongs to post-shot measures such as goals prevented. Read the chapter →
Why is a header worth less than a shot from the same spot?
Headers are scored much less often. In StatsBomb's older data, foot shots in the six-yard box were scored about 40% of the time and headers under 25%. Read the chapter →
Chapter 3 · Typical xG values
What is the xG of a penalty?
About 0.76 to 0.79, depending on the provider: Opta uses 0.79, StatsBomb 0.76 (0.78 in its 2022 update), Wyscout 0.76. Read the chapter →
What is the xG of a free kick?
Direct free kicks average around 0.05 to 0.06 xG in published Opta-based figures, roughly one goal in 17 to 20 attempts. Read the chapter →
What is a big chance in xG?
A big chance is an analyst's tag for a situation where a player should reasonably score. In older Opta data big chances averaged about 0.38 xG; Opta dropped the tag from its model in 2022. Read the chapter →
Chapter 4 · Why providers disagree
Why does Understat show different xG from FotMob?
They use different models: Understat its own neural network, FotMob Opta's. Different inputs and training data give different values for the same shots. Read the chapter →
Which xG is the most accurate?
No commercial provider publishes its accuracy figures, so no one can say for sure. Models with more information about each shot (keeper and defender positions) should do better in principle. Read the chapter →
Is there an official xG?
No. There is no FIFA, UEFA or league standard. Leagues and broadcasters show their data partner's model. Read the chapter →
Chapter 5 · Live xG
How often is live xG updated?
It depends on the provider. Xgoals' live xG updates about once a minute; premium feeds are faster. Read the chapter →
Why did the xG change after the match?
Live data is notated in real time and checked afterwards from video. The final data can move a shot's location or type, and so its xG. Read the chapter →
Why is there no xG at the start of a match?
On Xgoals' feed the first xG value usually arrives around the 11th minute. Before that it is unknown, not zero. Read the chapter →
Chapter 6 · xG myths
Does xG show who deserved to win?
No. It measures chance quality. The result also depends on finishing, goalkeeping and the small margins of single moments, and on how the score changed the way teams played. Read the chapter →
If a striker is below his xG, will he score soon?
Not because of it. Each shot is independent; past misses do not raise the chance of the next shot going in. Read the chapter →
Chapter 7 · Shot-value explorer
Is this the real xG of my shot?
No. It is a simple teaching model fitted to published averages. Real models know where the keeper and defenders are and what came before the shot. Read the chapter →
Why does a header score so much less?
Headers are harder to direct and to strike hard. In published data they go in at roughly half the rate of foot shots from the same place. Read the chapter →
Chapter 8 · xG calculator
Why Poisson?
Goals are rare and fairly independent, which is what a Poisson count describes. It is the standard first model in football analysis. Read the chapter →
Can I beat the bookmaker with this?
No calculator does that. Prices already hold the same information and more; this page shows what numbers mean, not what to bet. Read the chapter →
Chapter 9 · Shot map and xG race
Why are circles different sizes on a shot map?
The area of each circle shows the shot's xG, so the biggest circles are the best chances. Read the chapter →
What does a flat line in an xG race mean?
The team created nothing in that spell. It often follows a goal, when the leading side sits back. Read the chapter →
Chapter 10 · One big chance or many small
Is it better to have one big chance or many small ones?
For scoring at least once, one big chance is better. For scoring several goals, many chances give more room. Totals alone cannot tell them apart. Read the chapter →
What is shot quality?
xG divided by shots: the average value of a team's or player's attempts. Read the chapter →
Chapter 11 · xG for, against and difference
What is a good xGD?
Above zero means a team creates more than it allows. Title contenders usually sit well above +1 per match; relegation sides below zero. Read the chapter →
Is xGD better than goal difference?
For judging strength and predicting what comes next, usually yes, especially early in a season. Goal difference decides the table. Read the chapter →
Chapter 12 · Expected points
How is xPts calculated?
From the chance of a win and a draw: xPts = 3 × P(win) + P(draw). Those chances come from simulating the match's shots or computing their combinations exactly. Read the chapter →
Is an xPts table better than the real table?
It is a different view: the points a team's chances usually bring. It shows where results ran above or below the chances, but the real table decides titles. Read the chapter →
Chapter 13 · Sample size
How many matches before xG is reliable?
There is no single number. After about ten matches a trend shows; over a season xG is a good guide to a team's level, but it never removes chance. Read the chapter →
Why do teams over- or under-perform their xG for a whole season?
Goals are rare and every shot is a draw. Over 38 matches the spread is still about a third of a goal a match around the chances, a dozen goals or more in a season. Read the chapter →
Chapter 14 · Hit rate and base rate
Is a 50% hit rate good?
Only if the base rate is lower. If teams score again half the time anyway, a 50% signal adds nothing. Read the chapter →
Why not show only the alerts that worked?
Because then any signal looks good. A record is only worth something if it keeps every case. Read the chapter →
Chapter 15 · Finishing over xG
Is overperforming xG a skill?
Rarely, and it takes many seasons to show. In most cases a season above xG is followed by one closer to it. Read the chapter →
Will a team below its xG start scoring?
Its future scoring depends on the chances it keeps creating, not on the goals it missed. Each shot is independent. Read the chapter →
Chapter 16 · Game state
Do xG models know the score?
Most do not. That is why a team that leads early can end with less xG than a side that chased the game. Read the chapter →
What is game state?
The score at a given moment, from one team's point of view: leading, level or trailing. Read the chapter →
Chapter 17 · How accurate is xG?
Is xG accurate?
Good models are well calibrated over many shots: shots valued at 0.20 go in about 20% of the time. No provider publishes its exact figures. Read the chapter →
What is a Brier score?
The average squared difference between predicted probabilities and outcomes. For xG, good models score about 0.08; lower is better. Read the chapter →
Chapter 18 · xGOT and post-shot xG
Is xGOT always higher than xG?
No. A shot that misses the target has xGOT of zero, and a weak shot at the keeper can be worth less than its xG. Strong, well-placed shots raise it. Read the chapter →
Why do goalkeepers' stats use xGOT?
Because a keeper can only save what is on target. Comparing goals conceded with the xGOT of the shots faced shows how much the keeper added. Read the chapter →
Chapter 19 · Goals prevented
What is a good goals prevented figure?
Above zero means better than an average keeper on the same shots. Over a season, a few goals either way is common; consistent figures over several seasons say more. Read the chapter →
Is goals prevented the same as PSxG minus goals?
Yes. PSxG−GA, used by FBref with Opta's data, is the same idea under another name. Read the chapter →
Chapter 20 · Expected assists
What is the difference between xA and xAG?
Opta's xA values every completed pass by its chance of becoming an assist. xAG gives the passer the xG of the shot that followed; no shot, no credit. Read the chapter →
Is xA better than assists?
It shows the chances a player creates without depending on his teammates' finishing, so it is steadier from season to season. Read the chapter →
Chapter 21 · xGChain and xGBuildup
What is the difference between xGChain and xGBuildup?
xGChain credits everyone in a move that ends in a shot; xGBuildup leaves out the shooter and the last passer. Read the chapter →
Chapter 22 · Expected threat (xT)
Who invented expected threat?
Karun Singh popularised it in an interactive article; it values zones on a 16 by 12 grid by the chance of a goal following. Read the chapter →
Chapter 23 · OBV and VAEP
Is OBV the same as xG?
No. xG values shots; OBV and VAEP value every on-ball action by its effect on the chances of scoring and conceding. Read the chapter →
Chapter 24 · PPDA and field tilt
What is a good PPDA?
Lower means more pressing. Values are only comparable within one source, because the pressing zone and the actions counted vary. Read the chapter →
Does high field tilt mean the better team?
No. It shows where the game was played, not how good the chances were. Read the chapter →
Chapter 25 · Set pieces
Does npxG exclude corners?
No. npxG removes only penalties. Corners and free kicks are still included. Read the chapter →
What is the xG of a corner?
A corner itself has no xG; the shot that follows it gets one, depending on where and how it is taken. Read the chapter →
Chapter 26 · Women's football xG
Is xG the same for women's football?
The idea is the same; the values come from a separate model at Opta, trained on women's matches. Read the chapter →
Chapter 27 · History of xG
Who invented xG?
The idea grew from academic work (Pollard and Reep, 1997) and earlier hand notation by Charles Reep; Opta credits Sam Green with introducing its xG in 2012. Read the chapter →
When did xG appear on TV?
BBC's Match of the Day began showing xG in the 2017-18 season. Read the chapter →
Chapter 28 · xG and odds
Can xG help me win bets?
Bookmakers already use xG-type data, and research found their odds better calibrated than a simple xG model. xG helps you understand matches; it is not a betting edge. Read the chapter →
Is a team below its xG due a goal?
No. Each shot is independent; past misses do not raise the chance of the next shot going in. Read the chapter →
xG explains football; it is not a betting tip. 18+ · Please gamble responsibly.