Premier League Teams: Improvements or Drop-offs by Phase [2025/26 season]

Tracking improvements and drop-offs in key metrics from phase one (GW 1-13) to phase two (GW14-26) of the season so far.

Written on 13/02/2026

A Premier League campaign consists of 38 gameweeks, so after 26 rounds of games, we are officially two-thirds of the way through the season, with an FA Cup weekend providing a brief pause before it all resumes next week.

This provides a good opportunity to examine teams’ underlying numbers and assess how much they have improved or declined by comparing their metrics from Gameweeks 1–13 with those from Gameweeks 14–26. The first phase spans August to the end of November (GW 1-13), while the second phase begins in December, covering the festive period and the preceding midweek schedule (GW 14-26).

To conclude, we will analyse which players are high for expected goals involvement, and shot-ending possession chains in each phase (xGChain), highlighting the extent of overlap among the top performers in both phases.

To start off, let’s look at the improvement or drop-off in points and expected points in the second phase compared to the first.

xPoints (per Understat model) uses each match’s xG for and against, simulates the game many times using those shot probabilities, calculates the chances of win/draw/loss, and converts those probabilities into expected points.

Burnley have improved their xPoints, which was not a difficult feat given their poor start, but their actual points total has declined in the second phase. They are one of three anomalies in the table, alongside Manchester City and Brentford, who have both improved in actual points despite a drop in xPoints.

Tottenham are on the Burnley train as well, but from a slightly higher starting position in the first phase.

What this suggests is that when teams improve their underlying performance metrics, their xPoints tend to rise, which in turn is likely to translate into improved actual points over time. The reverse is generally true as well.

Next up, we look at goals scored and goals conceded to assess how teams have improved or declined.

Arsenal have gone from conceding seven goals in their first 13 games to 11 in the next 13, indicating a notable drop-off in the goals conceded column. Liverpool, Everton, Leeds United, West Ham United, Bournemouth and Manchester United are the sides that have shown improvement in both metrics, while Wolves have also done so from a lower starting point (they scored seven and conceded 28 in the first 13 games, bar cannot be any lower than that).

Brighton, Sunderland and Chelsea (the latter only by a smidgen in goals scored) have all dropped off in both metrics to some degree, while Crystal Palace have fallen off a cliff. Oliver Glasner’s side scored 17 and conceded 11 in their first 13 games, but that has shifted to 11 scored and 23 conceded in the last 13. A proper drop-off!

Next up, we turn to expected goals created and conceded. Can we see similarities to the actual goal metrics covered in the previous table?

The rest of this article is for paid subscribers (only £3.50 to access at least six data-driven breakdowns, player and team analysis, scout reports per month; £4 off for an annual subscription).