Goals are a noisy measure of attacking quality. A team can outplay an opponent completely and lose 1-0 because of a keeper error and a wonder strike. Expected goals (xG) is designed to cut through that noise.

xG measures the quality of every shot taken in a match by assigning each attempt a probability of resulting in a goal, based on the characteristics of the shot. Add up all those probabilities and you get the xG total for a team in that match.

---

What xG measures

Every shot is assigned an xG value between 0 and 1 based on factors like:

A shot from six yards out after a cut-back across the box might carry an xG of 0.45 (roughly a 45% chance of scoring). A long-range effort from 35 yards might carry 0.03.

Sum all the shot xG values for a team in a match and you get their total xG for that game. A team with 2.3 xG in a match created enough quality chances that they should, on average, have scored around two or three goals.

---

Why xG is more useful than goals

Goals in any single match are heavily influenced by variance. Great saves, woodwork, and goalkeeping errors all affect the scoreline without changing how well the teams actually played.

xG strips some of that variance away. A team consistently generating 2+ xG per game is creating good opportunities regardless of whether the goals are going in yet. A team with low xG but good results is likely experiencing better-than-expected finishing, a pattern that tends to regress towards the average over time.

For betting purposes, this matters because bookmakers set prices partly on recent results. A team on a good run despite low xG may be overpriced as a favourite. A team generating high xG but conceding from low-quality moments may be better than their recent results suggest.

---

xG against (xGA)

xG against measures the quality of shots a team has conceded. A low xGA indicates strong defensive positioning and goalkeeping, not just a run of opponents missing chances. A high xGA indicates defensive vulnerability regardless of recent clean sheets.

Looking at both xG (attacking quality) and xGA (defensive exposure) gives a more complete picture of a team than goals scored and conceded alone.

---

xG and the BetSignals model

BetSignals does not use an external xG feed as a direct input to the model. Instead, the models generate their own expected goals estimates, known as signalGoals (sG) for each fixture from goals and shots data using a probabilistic scoring model. These internally computed expected goals drive the score distribution, which in turn generates win probabilities, BTTS probabilities, and goal totals.

The sG displayed on BetSignals is the model's own estimate of expected goals for each side, not a third-party xG figure.

---

Practical applications

Identifying overachieving and underachieving teams. A team with 2.0 sG per game but only 0.9 goals per game is likely unlucky in finishing terms. Their results may improve without any change in actual performance.

Assessing Over/Under markets. Combined sG across both teams in recent fixtures is a better predictor of whether a match will go over 2.5 goals than goals per game alone.

Evaluating goalkeepers. A keeper with goals against well below their xGA is performing above the baseline. This tends to regress, worth factoring in before backing a team for a clean sheet.

---

The limitations of xG

xG models vary between providers. The same shot might be assigned 0.25 by one model and 0.18 by another, depending on the input variables and training data. No xG model is definitive.

xG also does not account for the quality of the shooter, some players consistently overperform or underperform their xG due to finishing ability that the model cannot capture. Elite strikers often outperform xG reliably; average forwards often underperform it.

---

Next reads

---

18+ | Gambling should be enjoyable. If it stops being fun, take a break. For support and advice visit BeGambleAware.org. See our full responsible gambling guide.