Friday 7 September 2012

Team Page Update: Everton 6 week Points Projection

Here is my first stab at projecting points per player, starting out with the mighty Toffees.  There is quite a bit going behind the scenes, and it doesn't use the F.SCORE for projections, but a combination of both the club's and the player's 2011/12 season stats. adjusted  based on this season's performance so far, all based only on underlying stats, key passes and shots on targets rather than assists and goals.

For both players and the team as a whole there is a "forgetting factor" built in, which essentially reduces last season's influence on current/projected form with each successive match played.  It's going to need some heavy tweaking as we go on but I am preparing to be quite aggressive with the weighting,  biasing it toward recent results.  For some clubs I have not taken their entire 2011/12 season but a section, notably for Sunderland and QPR who improved under new management through the season,

Looking at the projections for Everton you can see the obviouls subjects are picked out to do well - so I ask myself what's the point?  Where's the value?  Well, firstly Everton have had a well documented good start to the season so their players have been highlighted, but not every team has been on the radar like Everton have.   One of my thoughts on FPL strategy is that it's the unheralded clubs and players that make the difference in the long run, assuming you captain Van Persie when everyone else does that is!   So, whilst highlighting that Baines, Jelavic and co. are going to do well is not so insightful, but when this is done for all 20 team's it's usefulness will become apparent in picking out the "hidden points".

Secondly, and perhaps in a slight contradictory note to the above, you can see that Pienaar has a better projection and underlying stats. than Fellaini but Fellaini's ownership is twice Pienaar's.  Fellaini's amazing performance in the game against United and his two goals have created such hype that his ownership has rocketed but I would have subjectively suggested that Pienaar is the better pick, and these projections back that up.

Thirdly, once completed for all team's I will be able to produce projections for my FPL squad and hope that will allow me to quickly see where improvements can be made.  

Out of interest I compared my projections to those on Fantasy Football Scout, which is a feature available to members.  The correlation was very positive (0.86) except for two outliers  - Hibbert & Jelavic.   FFS have Jelavic doing better than I have, and Hibbert doing worse.  I have probably given Hibbert more gametime than FFS, based on Gibson's groin strain, and I have not projected Jelavic to be as clinical as he was last year (9 goals from 17 shots on target, I expect more link 6 or 7 goals from 17 SoT).  Being an Everton fan hope I'm wrong!! 

So, having convinced myself, have I convinced you??  There'll be a more detailed post showing my workings soon but I'll be working on projections for the other 19 teams next.

Update 10-Sep:  Prediction model updated to reflect latest tinkering :)  In general a boost to FWDs. 

No comments:

Post a Comment