Points versus Expected
Points versus expected is expected points minus actual points, so a positive figure means a team has banked fewer points than their performances deserved.
Why it matters. It is the strongest regression signal in the table. Gaps this large rarely survive a full season in either direction.
Built from one subtraction
Points versus expected is expected points minus actual points. A positive figure means performances deserved more than the table currently shows. A negative figure means results have run ahead of the performances producing them.
That sign convention runs the opposite way to the player-level deltas on this site, such as xG difference or xA difference, which are actual minus expected. Here the table asks what was deserved and compares it against what arrived. At player level the question is reversed: what arrived, measured against what was deserved. Both are deliberate, and neither needs correcting to match the other.
Three ways the gap plays out
The team due an upturn. A side sits mid-table on points but carries a ΔPts of plus six or more. Their performances have consistently created and conceded chances at a level that should be earning more wins than the results show. That gap is the strongest single reason to expect their league position to climb over the coming weeks, even with no change in personnel or style.
The team due a comedown. The mirror case: a side in the top half with a ΔPts running negative by a similar margin. Results have been winning matches their underlying play did not deserve, often through a run of converted low-quality chances or a favourable set of late goals. Their position looks stronger than their level of performance.
Settling the form argument. A team goes on a run of five wins and the question becomes whether the form is real. Checking ΔPts across that run answers it directly. A run built on genuinely improved chance quality shows a ΔPts close to zero or trending positive during it. A run built on results alone shows the actual points pulling well clear of expected points over the same stretch. That is the tell that the level of play has not actually caught up with the scoreline.
The strongest regression signal on the table
Of every gap on the expected table, ΔPts moves the most reliably back towards zero. Points already compress a season's worth of variance into a single number. Once a gap this large opens up, it is describing a real and sustained mismatch between performance and result rather than one unlucky afternoon. That is what makes it worth checking before trusting a current league position at face value.
Where the read goes wrong
A large gap does not always close within the same season. A gap spotted with five matches left has less time to correct than one spotted after ten games. Some gaps carry over into the following campaign instead of resolving inside the current one.
A gap closing does not mean it closes evenly. The correction can arrive in a short burst, several matches where results and performances finally line up, rather than a steady week by week drift. Do not expect the gap to shrink by the same amount every round.
Common questions
- Which way round does this delta run?
- Expected minus actual. A positive number means the results undersell the performances. The player deltas use the opposite convention, actual minus expected, and the two are not meant to be harmonised.
- Does a large gap always close?
- Not always, but it narrows far more often than it widens. Late in a season a persistent gap is better read as a description of how the campaign has gone than as a forecast of the run in.
- What counts as a large gap worth paying attention to?
- Anything above roughly four or five points either way, ten matches or more into a season. Below that, normal match to match variance in win probability is enough to explain it on its own.
- How is expected points itself calculated?
- Each match's expected goals for both teams feed a model that estimates win, draw and defeat probabilities, which are then weighted by their point values and summed across the season. See expected points for the full method.
- Can a negative ΔPts mean a team is actually well run?
- Sometimes. A team that consistently manages leads well, or finishes matches strongly with substitutes, can post a real repeatable edge over its expected points. That is rarer than it sounds, and most negative gaps do fade.
- Should ΔPts be trusted more early or late in a season?
- Late, once there are enough matches for the swings to average out. A large gap after three or four games is mostly noise. The same gap after twenty is a genuine pattern.
Related terms
See ΔPts across the Premier League
Every player, team and match measured on the chances created rather than the scoreline.
Last updated