xG, form and H2H — the statistics that actually matter
9 min read
Every match comes with a dozen numbers. Some change how you read the fixture, some are filler, and a few mislead more than their absence would. Here is how to tell them apart.
xG — the best single metric, if you know what it measures
Expected goals assign each chance a scoring probability based on where and how the shot was taken. The match total says how many goals “should” have come from those chances.
Its value lies in being far less random than the scoreline. A team losing 0-1 with 2.4 xG to 0.3 played better, and would win most reruns of that match. The table remembers the result, but xG predicts the next fixtures better.
Three caveats. It ignores who was shooting — every player is treated identically. High xG from one chance (a penalty is roughly 0.76) is not the same as identical xG from ten shots. And below a dozen or so matches, xG differences sit inside the noise.
Form — five matches is not enough
The W-W-D-W-L strip is everywhere and means almost nothing without context. Across five games, a run of draws and a run of wins are often separated by one shot off the bar.
Read form with three questions:
- Against whom? Three wins over the bottom half and three defeats to the top look the same on the strip and mean the opposite.
- Where? Home and away records can differ more than the records of two different clubs.
- How? Wins after dominance and wins from one counter-attack look identical in the table and nothing alike in the shot data.
H2H — the most overrated number in football
Head-to-head records seduce because they read as a story: “this side never suits them”. But matches from four years ago involved different squads, different managers and different form. Football turns over almost completely every few seasons.
H2H is useful in two cases: derbies, where the emotional context genuinely persists, and pairs of clubs with long-standing contrasting styles whose meetings keep producing the same script — low-scoring games, for instance. Otherwise limit it to the last three years and treat it as trivia.
Underrated and worth attention
- Which players are missing, not how many. Losing the first-choice keeper or playmaker outweighs three absent substitutes many times over.
- Fixture congestion. A third match in a week is a different team from the same eleven after a week off.
- A managerial change. Every statistic from before it describes a plan that no longer exists.
- A gap between league position and shot data. Sustained over a dozen matches, it usually closes one way or the other.
What to skip
- Possession. On its own it does not correlate with results — counter-attacking sides concede it deliberately.
- Corner counts. Sometimes a symptom of pressure, but not a route to goals by themselves.
- Fouls and cards, unless you are specifically looking at cards or the referee.
- “Streaks” such as consecutive matches with a goal. Usually coincidence described with the force of law.
Putting it together
Start with xG over a dozen or more matches to establish the real level of both sides. Adjust for absences and for a managerial change if there was one. Check both fixture lists for the past week. Only then look at form and H2H — as a test of whether anything contradicts the conclusion, not as its basis.
Our analyses follow the same order and state a data-quality assessment each time. If the season has just started and the sample is too small, the text says so — because a number computed from three matches looks exactly as authoritative as one from thirty, and means something entirely different.