xFPL: A beginner's guide
This is a brand new members series by @FPLBraveheart - a keen poker player who effectively applies the principles of variance and probability to FPL. With several top rank finishes to his name, Braveheart does to devastating effect. One of the main concepts he uses is xFPL - I'll leave him to explain what it is, why it's important…
Cracking read @braveheart. Great addition to the site, will look forward to these articles.
Really interesting article and great to read. Looking ford to the next one.
1. Would love to know how you calculate xCS
2. Some people would say that better players will perform above xG (hence they are better players). Not something I agree with, but what's your view?
3. When you mention about luck turning, you mean going back to normal rather than bad luck going good and good going to bad. So if I flicked a coin 10 times and got 8 heads through good luck, if tossing another ten times my luck should turn and I expect 5 heads (not 2).
4. Do you make any adjustments for quality of opponents or game state. By game state I mean if a team goes 1-0 they become less likely to score the next goal and players xG dips. Of course being 1-0 down you chase the game more and so xG increases.
No need to answer now (as each is probably quite complex and long), though if the opportunity arises you could slip in an answer in future articles in the coming weeks and month, which I will definitely be reading as this is excellent stuff.
Great article and an exciting series that I'm looking forward to tracking, love stuff like this! An observation though; having been tracking xStats for various players lately, I was surprised not to see certain players on your list (e.g. Doherty). I looked at the Google Drive sheet and noticed a comment on there stating that defenders were restricted to >2% owned (and midfielders >1% owned). Could I suggest lifting these restrictions as it could be skewing the top 10, where a low owned player is actually a real gem with lots of potential? Also, out of interest, over what period do you calculate a players expected performance; I assume it's since the start of the season at the moment? Many thanks!
Carl, thank you for your comments. I think I will respond to your first two questions in future articles. In particular, I'm keen to answer your second question in my "talking point" section in next week's article (or a future article if some major talking point arises this weekend). I think it is an interesting question and worth looking into (with reference to previous season stats).
3. Of course, the luck should go back to normal. It would be comparatively bad however when you compare it with that player's running good period with the first ten coin flicks. I perhaps tried too hard and oversimplified variance! Apologies!
4. The spreadsheet makes no adjustments for quality of opponents or the game state. I previously toyed with tying the calculation with the Official FPL Fixture Difficulty Rating. However, I felt combining an analytical stat with a predictive stat did not seem to blend well. Opta's xG stats model takes into account the positioning of the goalkeeper and the quality of the goalkeeper. The stats I use due likewise. I'd be keen for any ideas that the Hub Community have to improve the spreadsheet. I want to make it as good a resource as it can be so any input you have would be most welcome.
Thanks for reaching out once again. I hope my responses were helpful.
TriXaM, thank you for reaching out and for your input about how we can improve the spreadsheet. Those restrictions were put in place to weed out a lot of the 3.9m defenders who have no hope of ever playing (therefore making compiling the spreadsheet easier and more efficient for me). However, I agree that 1% and 2% might be too high as it omits certain players who may be uber differentials (like Doherty as you mentioned). Doherty's xFPL score was: 15.2 and his value is: 3.45. That makes him the second top defender behind Alonso in terms of xFPL and the top defender in terms of xValue. This is a very strong argument for widening my parameters and so I will do that for future weeks!
The stats are currently for the first five gameweeks but once we get beyond GW6, I'll be looking at the last 6 GWs.
It depends if you have other areas you'd be better off prioritising. If you don't, and have decided to ditch Mendy, Robertson is an excellent shout. When paying 6m+ for a defender what you should be looking for is attacking potential. The long term CS potential should be taken as read. Robertson has high prospects of returning points even when Liverpool don't keep a CS. So don't be scared of his fixtures!