The lucky and the good (expected goals)

This is the first in a new regular feature, where I give an overview of players who have over/under achieved based on expected goals.

Overview expected goals (xg)

Analysis of expected goals is now becoming an important new way of analysing performance that is increasingly being taken notice of by clubs, betting agencies and other parties interested in gaining a deeper understanding of true performance.

How does it work?

All shots are analysed based on a number of factors such as distance from goal, angle of shot, part of body the ball came off, type of pass that led to the shot etc. So a shot from close range is given a higher ranking than one from outside the box, while a shot in front of goal has a higher ranking than one from out wide.

How is it used?

It measures past performance and can show when teams or players have performed better or worse than mere results can explain. It can be a good indicator of luck, a team that loses One nil despite many chances compared to the opposition, goes down as a loss, however xg provides a truer picture, and is far more indicative of what should have happened.

Is it perfect?

No, however it is the best there currently is. No system is perfect. For example for the past Three seasons Harry Kane has far outscored the goals he was expected to score. This indicates he is simply very good, and performs better than expected because he is better than expected. Overall however it is a very good indication of how a team/player has performed.

Enough of the theory

No scatter graphs here, just snapshots of the most salient points from the season so far. tables which I shall update every week that will show at a glance possible outliers in terms of teams and players. Only the top Three from each category will be shown

League table Vs Expected league table

Early days yet, only Three matches in the book so probably too early to make too many assumptions, however the Three teams that the model suggests to be too high in the league look very plausible.  Swansea, Wba and Huddersfield all look to be in for some regression.

The surprise in the table is Newcastle, it looks as if they have been playing far better than their league position suggests. Tottenham we all know about, clearly they have been unlucky, you don’t need to be a stats guru to know that. If they were not high on the list then the model would have to be bought into question.

Expected goals, the over and under achievers

Tottenham are a side that normally score more than expected, so to see them under achieving is a big surprise and an indication that they have been very unlucky, they should have had Five or Six goals by now, and to see them on only Four does not mean they have been poor. Expect improvement, now does not appear to be the time to sell their assets, quite the opposite.

Southampton too should have had a few more goals, some of the stuff they missed was far easier to score than not. They may turn out to be one of those frustrating teams that constantly underperform statistics, but it is too soon to make that call. If you hold, keep.

Chelsea are an interesting one. They have scores Six times, but the shot models indicate they should have only scored 3.4 goals. We all know how good Chelsea can be, however in the first Three matches it looks like,they were flattered by their Six goals.

Goals conceded

We have all been applauding the Man U defence, however it looks as if the Chelsea defence has performed just as well, but have been unlucky to concede Four times. The last minute goal for Burnley did not help matters. West Ham should have only conceded Eight which is still terrible, but they should be in for some improvement at least.

Huddersfield, Wba and Swansea again turn up on a table which suggests they have over achieved. Those like myself who are considering Naughton, should maybe pause for thought. These statistics show they have been lucky so far, and instead of conceding Four could well have let in Seven by now.

Best players

Lukaku, Salah and Kane have been the Three standouts so far. The big story here is that it is a prime  example of how picking the right players does not always result in the most points. By rights Kane should have had at least a couple of goals by now, and him not scoring is more to do with luck than any August curse.

Expected assists is another category I have added although I have no idea if it is as accurate as for expected goals.  Robertson is a surprise at number Three, he sits just ahead of Eriksen who was in Fourth place. If Robertson can cement his place in the Liverpool side, he could be in for a high scoring one fpl wise.

Conclusion

With only Three weeks of data the rankings will probably be subject to rapid change. They seem to pass the eye test however, and have not thrown up a load of obscure randoms, they look in short, to have correctly separated the good from the lucky.

21 Comments

  1. Thankyou, it will be interesting to see how it progresses. If anyone has any requests for the stats of any particular player, just leave a comment here. I only included top Three but have them all

  2. Great read, Robertson is very interesting and should cement LB over Moreno over time. Would be interested in seeing Laca’s stats before the Bournemouth game 🙂 thanks

  3. cheers TL … this is really useful stuff. The Naughton one is interesting. Hazard coming back should sort out the Che expected goals!

  4. I was considering Naughton too mainly as a means of raising cash for hazard assuming he has some positive involvement against Greece. You’ve got me thinking now. This is a great feature though. Very insightful indeed.

    • Has to be remembered the samples are still very small, and it takes a while for league tables to settle down, so although Swansea 19th on the expected table it woupd not take much to see them improve

  5. Thanks for these, interesting content (as always) and hopefully useful as a predicter of future points

    Would be good to see Southampton expected goals conceded, and Pogba expected goals? Cheers

  6. Another thought, would be interesting to see tables for Top 10 expected goals and assists alongside each other, as players appearing on both lists surely good picks for FPL

Leave a Reply

Your email address will not be published.