Every defense concedes goals it probably shouldn't. A deflection wrong-foots the keeper, a shot from 30 yards finds the one corner he can't reach, a scuffed cross loops in off the crossbar. Over time, these moments even out. In the short term, though, they pile up - and the gap between goals conceded and expected goals against (xGA) tells you exactly where.
This is the defensive luck signal. A team conceding far more than its xGA is likely getting punished by finishing quality or misfortune beyond its structural defending. A team conceding far less is probably riding some combination of elite goalkeeping, opponent wastefulness, or plain good fortune. Three matchdays into the Premier League season, the spread is already striking. If you've been scanning the Premier League xG table looking for teams whose results don't match reality, this is where to start.
Most Cursed by Defensive Luck
1. Crystal Palace - 8 goals conceded against 3.3 xGA. That's a gap of 4.7 goals, comfortably the worst in the league. Palace are defending like a mid-table side and conceding like a side heading for the trap door. The quality of chances they've allowed doesn't remotely justify the damage on the scoreboard.
2. Ipswich - 8 conceded vs 5.2 xGA. Ipswich have genuinely been leaky, but even their actual defensive structure only accounts for about five of those goals. The other three arrived from somewhere less explicable.
3. Fulham - 7 conceded vs 4.3 xGA. A 2.7-goal gap is significant this early. Fulham's underlying numbers suggest a defense performing closer to average. Their results table disagrees.
4. Coventry - 5 conceded vs 2.6 xGA. Newly promoted and already nearly doubling their expected goals against. The margins at this level are thin enough without the ball finding the net every time it sniffs the box.
5. Chelsea - 7 conceded vs 4.8 xGA. A 2.2-goal gap won't spark panic at Stamford Bridge, but it's enough to turn a reasonable defensive start into a genuinely poor one on paper.
Most Blessed by Defensive Luck
1. Hull City - 0 conceded against 3.1 xGA. Zero. Three matches, 3.1 expected goals worth of chances created against them, and not a single one has gone in. This is the kind of number that does not survive contact with October.
2. Everton - 3 conceded vs 4.8 xGA. A 1.8-goal gap in their favor, which for Everton qualifies as a minor miracle in any department.
3. Leeds - 2 conceded vs 2.3 xGA. A tiny gap, barely worth noting, except that it places them third on this list. Leeds are defending well and getting results that roughly match.
4. Newcastle - 4 conceded vs 4.1 xGA. Almost perfectly calibrated. A 0.1-goal gap is essentially noise.
5. Brentford - 2 conceded vs 2.0 xGA. The gold standard of what-you-see-is-what-you-get defending. Brentford have been exactly as good as they look.
What This Tells Us
The range from Crystal Palace (+4.7 goals conceded above expectation) to Hull City (-3.1 below) spans nearly eight goals. In three matches. That's an enormous band of variance, and historically, almost all of it corrects.
Defensive luck is one of the fastest-regressing signals in football analytics. Goalkeepers don't permanently save everything, and they don't permanently let everything in. Opponents don't keep finishing at 150% of expected conversion, and they don't keep missing every decent look they get. The research consistently shows that goals-minus-xGA converges toward zero faster than almost any other performance gap.
For Palace, that's encouraging. Their actual defending hasn't been 8-goals-in-three-matches bad. For Hull City, the news is less welcome.
Three clean sheets against 3.1 xGA is a story. It's just not a story that tends to have a long second chapter.
