The statistical tendency for extreme performances — very high or very low — to move toward average over time. A core principle behind xG-based betting.
Regression to the mean is a statistical phenomenon where an unusually high or low performance in one period tends to be followed by a performance closer to average in the next. In football, a striker who scores 8 goals from 3 xG in the first month of the season is almost certainly going to score fewer goals per xG going forward — not because their ability has changed, but because they were running above expected conversion rates.
Similarly, a team that concedes 5 goals from just 1.8 xGA is likely to concede fewer goals in subsequent games as goalkeeper and defender luck normalises. Regression to the mean is not a guarantee — it is a probability-weighted expectation over a sample of games.
Regression to the mean is one of the most exploitable inefficiencies in football betting markets. Bookmakers and public bettors frequently overweight recent results (goals scored, goals conceded) without adjusting for the underlying xG. A team conceding 8 goals in 3 games from 4.5 xGA will be priced up as a soft defensive side — but the xG tells you they have been unlucky and are likely to tighten defensively.
The xG divergence model (actual goals minus xG) is the primary tool for identifying regression candidates. A team with -0.8 xG divergence per game over 5 games is almost certainly running cold and due for improvement. Betting on them before the market catches up is a classic regression trade.
xG (Expected Goals)
A metric that scores every shot by its probability of resulting in a goal, based on factors like shot location, angle, and assist type.
xGA (Expected Goals Against)
The expected goals conceded by a team — a measure of defensive quality based on the quality of chances allowed, not just goals shipped.
npxG (Non-Penalty Expected Goals)
Expected goals with penalties removed — a purer measure of a team or player's open-play goalscoring threat.
Value Betting
Betting at odds that are higher than the true probability of the outcome — finding bets where the bookmaker has underestimated the chances of an event.
Market Efficiency
The degree to which bookmaker odds already reflect all available information — highly efficient markets offer less value; inefficient markets offer more exploitable edges.
For informational and educational purposes only. Disclaimer
