I think you're missing a couple of key factors.
We are outperforming our xG, outperforming our xGA, yet underperforming our xPts. That's actually a pretty unusual combination; what does that mean? Lets look at those one at a time:
xG: we have 55 goals from 46.3 xG, ie over-performed by 8.7.
Insofar as shooting is a skill, but each individual shot has an element of luck, then we can say that this net gain of +8.7 goals is a combination of luck and skill. If you look at the underlying stats, we're ahead by a bit over 1 standard deviation, which means it's probably not just luck. Especially when you consider that we've beaten our xG for 5 seasons in a row. Our recruitment of goal-scorers is such that we actually expect to beat our xG each season.
xGA: we have conceded 41 goals from an xG of 47.4, ie over-performed by 6.4.
Insofar as goalkeeping is a skill (as well as blocking / pressuring by defenders), we can sat this this net gain of 6.4 goals against is a combination of luck and skill. Again looking at the underlying stats, we are ahead by about 1.8 standard deviations, which is enough to say that there is almost certainly a large component of skill in that number, ie Vicario is an elite shot-stopper.
xPts: we have collected 34 points vs an expected 38.4, ie under-performed by 4.4.
What does that mean? Well basically the Opta xPts calculation looks at the xG for and against of each game, and calculates the probability of win / draw / loss based on that xG, assuming
average quality of strikers and average goalkeeping from both teams. If you take into account our excellent shooting and goalkeeping, the xPts would actually be 45.8 instead of 38.4.
So how can we underperform our xPts while simultaneously overperforming on shooting and goalkeeping? The answer is around luck / timing. In the games where we have under-performed our xG, or have conceded more goals than our xGA, it has almost invariably cost us points. And in the games where we have over-performed our xG, that was often just the difference between winning by 1 goal or winning by 2 or 3.
So in fact, as we get close to your "magical sample size", assuming we continue to employ better-than-average goal-scorers and better-than-average goalkeepers, we should expect to:
- Continue to out-perform our xG
- Continue to out-perform out xGA
- Due to (1) and (2), start to outperform our xPts
In fact that's finally started to happen in the last 5 games or so, particularly that recent run of 3 wins (Brentford, United, Ipswich) where we got 9 points from an xPts of 5.0. But before that, Newcastle away, Arse home, Brighton away, Palace away, Ipswich home, Chelsea home, Forest away, Wolves home, Arse away, Everton away, Leicester home, City home were all games that were close enough on xG to have a decent chance of collecting points, but where our finishing let us down (or opponent's finishing undid us). We had only really had comparable luck go in our favour in City away and Villa home prior to that recent run of 3.
Call it excuses, copium, whatever, but it's really just stats, and the stats suggest that things are likely to get better, not worse, as the sample size gets bigger.
As an aside: those much-referenced first 10 matches last season, 26 Pts, xPts (adjusted for shooting and keeping skill): 18.3. So swings and roundabouts, sometimes it works for you, sometimes against.
