Expected Points
Expected points estimate the points a team's performances deserved, by turning each match's expected goals into win, draw and loss probabilities.
Why it matters. Set against actual points it is the engine behind the whole expected table, and the clearest measure of how kind or cruel a season has been.
From chance quality to match points
Points come from results, not goals, and results are not a straight line from expected goals. A team can create 2.5 xG and still lose if the finishing goes against them. That 2.5 xG performance and a 0.4 xG performance are worth wildly different amounts in the league table, even though both can end in a defeat.
Expected points closes that gap in three steps.
- Model the match. Both teams' xG for that game feeds a model that estimates the probability of a win, a draw and a defeat for each side.
- Weight the outcomes. Those probabilities are multiplied by what each result is worth: three points for a win, one for a draw, none for a defeat.
- Add it up. Sum the weighted values across every match played, and the result is a running total in the same units as the real table.
A team that should have won two thirds of its matches, on the balance of play, ends up with an expected points total that reflects that. It makes no difference what their actual results said.
xPts rewrites three verdicts
The same real points, two different stories. Two teams sit level on points midway through a season. One has an xPts total well above their actual points: the underlying performances have been stronger than the results. The other's xPts sits below their actual points: results have been flattering them. The table treats them as equals. The underlying numbers do not.
A trend, not just a snapshot. Track xPts over a rolling run of matches rather than the season total alone. An upturn or a slump often shows in the underlying numbers before it shows in the results. A team whose recent performances have improved but whose points have not yet followed is a pattern worth watching.
The newly promoted side starting hot. A promoted team wins its first five matches while creating the chances of a relegation candidate. Their real points are excellent. Their xPts says the level of performance behind those results is not sustainable at this level, and results typically correct towards it as the season wears on.
Where the arguments break down
"It is a prediction of the final table." It describes matches already played, not matches still to come. Over and under performance both tend to fade, so it correlates with what happens next. It still makes no claim about future fixtures, injuries or form.
"A team with a high xPts is guaranteed to move up the table." Guaranteed is too strong a word for anything in football. It says the underlying performances deserved more than the table currently shows. That tends to correct over a large enough sample, not on any specific run of results.
"It double counts luck by building on xG, which is also luck-based." xG is built from where and how a shot was taken, independent of what the scoreline ends up being. There is nothing circular about it. Expected points takes an already independent measure of chance quality and converts it into the same units the table uses, nothing more.
Most expected points this season
Common questions
- How are expected points calculated?
- For each match, both teams' expected goals feed a model that estimates the chance of a win, a draw and a defeat. Those probabilities are weighted by their point values, a win worth three and a draw worth one, then summed across the season.
- Why not simply compare expected goal difference?
- Because football does not pay out in goals. A team that wins narrowly several times and loses heavily once can have a modest goal difference and a healthy points tally. Expected points convert chance quality into the currency the table is actually kept in.
- Is expected points a prediction of where a team will finish?
- No. It describes what already happened, converted into a fairer scoreline. It is correlated with what tends to come next, because underperformance and overperformance both tend to fade, but it is not a forecast in its own right.
- Can a team have a great expected points record and still get relegated?
- Yes, and it happens most seasons somewhere in the league. A team can create the chances of a mid-table side and still lose the moments that decide matches. They finish with fewer points than their performances deserved, however the table reads at the end.
- Does expected points use the whole match, or just the final score?
- The whole match. It is built from the chance quality both teams created, shot by shot, not from the scoreline. Two matches that both finish 1-0 can produce very different expected points contributions depending on how one-sided the underlying play actually was.
Related terms
See xPts across the Premier League
Every player, team and match measured on the chances created rather than the scoreline.
Last updated