The Premier League table tells one story. The underlying data tells a very different one, and the gap between the two has rarely been this wide this late in a season.
Wolves have accumulated 15.19 fewer points than their expected-points model suggests they deserve. Aston Villa have banked 15.79 more than theirs says they should. That is a 31-point swing between two clubs operating in the same league, under the same rules, subject to the same physics. One of those physics is regression.
Due a Correction
Wolves are the Premier League's most aggressively unlucky side by a considerable margin. A Luck Index of -53 is the kind of number that makes modelers double-check their inputs. Their net expected goal difference sits at -23.57, which means, yes, they have genuinely been a poor side. But poor and this poor are different conversations. A 15.19 xPoints gap means results have run colder than even a bad underlying process would predict. Markets pricing Wolves as a bottom-feeder in every remaining fixture may not be fully accounting for the degree to which finishing variance and save percentages have conspired against them. Teams carrying gaps this large at this stage of the season tend to see partial correction over the final 5-8 matches, even if they remain fundamentally flawed.
Chelsea present a more intriguing case. Their net xG is a healthy +17.95, a figure that belongs to a top-four contender, not a club sitting 6.50 points below its expected total. A Luck Index of -43 suggests that Chelsea have been generating quality chances and conceding relatively few, then watching the scoreboard disagree. Historically, sides with this profile and this much runway left tend to go on visible points surges. The underlying process is strong. The finishing has simply been unkind.
Crystal Palace round out the underperformers with a -32 Luck Index and a 5.07 xPoints deficit. Their net xG of +7.64 marks them as a side that has, on balance, been winning the territorial and chance-creation battle more often than not. The table does not reflect this. Bookmakers tend to adjust lines slowly when a mid-table club's results lag its process by this amount.
Living on Borrowed Luck
Aston Villa top the overperformance chart with a Luck Index of 58 and an xPoints gap of +15.79, all while carrying a net xG of -5.51. Read that again. Villa have been outshot on an expected-goals basis across the season and yet sit nearly 16 points above where the model places them. That is an extraordinary amount of variance running in one direction. Variance like this typically regresses, and when it does, it tends to arrive in clusters.
Manchester City carry a Luck Index of 52 and a 6.27 xPoints surplus. Unlike Villa, their net xG of +12.17 confirms genuine quality. But genuine quality plus six extra points is still six extra points. City have been good and lucky, which is a combination the standings love and the models flag.
Manchester United share City's exact xPoints gap (+6.27) and net xG (+12.17), an oddity worth noting. Their Luck Index of 34 is lower than City's, suggesting the overperformance is less extreme but still meaningful. United's underlying numbers support a strong side. Just not quite as strong as their current points haul implies.
Sometimes the standings just lie for a while.
The Regression Window
In Premier League seasons, xPoints gaps of this magnitude tend to close by 40-60% over a club's next 5-10 matches. Full correction rarely happens in a single run of fixtures, but partial correction is one of the most reliable phenomena in football analytics. The rate of closure accelerates when the underlying process (net xG) is clearly misaligned with results, as it is for several of the clubs listed above. None of this is a guarantee. It is a probability distribution, and right now, that distribution is tilted.