Here’s a slightly different article from normal where I look at a different way of identifying potential in midfielders for the coming season. I’m no mathematician but it does giving some interesting slant on your picks for the season. It’s not meant to identify the best midfielders just those who have the potential to score above expectations
fantasy premier league midfielders – looking at regression as a way of picking players
If you read alot of the statistical articles you’ll see that the general theory is that shots on target is the best way of predicting goals. However if it was that predictable I would be in Monaco sunning myself rather than Manchester so just take it as something else to factor into your decision making. Just a reminder that again it’s not a way of predicting the highest scorer just point to those who could score goals above expectations. What we are looking for is to find someone who was unlucky last season and this season should score at a ratio more in line with their historic expectation. It’s just sorting out the bad luck from those that changed for other reasons such as positional change.
Having taken the shots on target as a predictor we’ve looked to find midfielders with the following collective criteria:
- See if there are players who last season scored from shots on target that was at a ratio considerably worse than the average ratio of midfielders.
- We then looked to see if that was different than their ratio of the previous season to just check that the ratio last season isn’t just their normal.
- For my money it’s preferable that person is someone who plays in an advanced midfield position rather than a deeper midfielder who consistently shoots from distance
- Lastly that player should also have a history of a minimum number of shots. No point in having a theory that gets 2 goals over the season
The theory being that you identify a player who last year had a shots on target ratio lower than the midfield average and their ratio last year. Therefore the theory is that they will return to their average ratio having been unlucky last season. That player should be an advanced midfielder rather than a CM and they should have a minimum of shots not someone who will score 1 or 2 goals a season. Again it’s differentiating luck from a change in circumstances or other reasons for the drop
So here’s a table which ticks off the criteria 1 by 1. If they fail one of the criteria above then they are given a N in the box and they are dropped from consideration. Hopefully it will be clear!
So out of the 6 midfielders who last season who had ratio of shots on target to goals lower than the rest of the midfield average we quickly eliminated Lamela, Coutinho and Barkley. 2 of which will shoot from distance and it’s a normal ratio for them so no reason to assume it will improve.
That left Ramsey, Toure and Eriksen who had ratio’s worse than the 14/15 season so at least a chance they could return to that 14/15 seasons better ratio this season coming. However I eliminated Ramsey and Toure as my preference for midfielders is to have advanced midfielders rather than CMs who rely on box to box movement and shots from distance.
That left 1 midfielder, Spurs Christian Eriksen. I double checked his 13/14 season SOT/goals ratio just to see what that showed and that also showed a lower ratio than last season. That again gives hope that he can return to his better normal ratios this season. Helpfully the amount of shots on target each season is consistent too. The problem with all this obviously is working out when the goals will come as it’s unlikely you’ll hold him all season. However if they are a consistent rather than a streaky scorer that will help
So a brief look at Eriksen
Here’s his key stats from last season:
So not bad at all. Consistent (although this includes assists) but with a strong 10+ pts % and finished the season in good form. He also creates chances and gets assists plus has good set piece representation. Apart from that he also had a rest over the summer with Denmark not in the Euros.
What about fixtures:
Everton (A)
Crystal Palace (H)
Liverpool (H)
Stoke (A)
Sunderland (H)
Middlesbrough (A)
Could be worse.
He’s in my team currently.
So what if I did this before the 15/16 season?
So here’s the analysis I would have done before last season. What would it have produced?
Here’s the answer.
So after you stop laughing Sterling did return to his historic ratio he just lost his place so his number of shots on target just reduced drastically as he only played just over 56% of the possible minutes. Before that he did qualify for the number of shots posting a decent amount per year. I’m not a fan though.
Coutinho and Bolasie will always have low ratios due to their desire to shoot from distance.
The interesting point is that of Mahrez in particular. He performed well under the average ratio but posted a decent number of shots on target. Last season he increased his shot number by 50% but also his conversion ratio improved to a typical midfield conversion ratio.
Tadic also improved his to a more typical ratio and posted the same amount of shots on target so you could also say he also would have been a good pick.
So who are the leading shots on target midfielders
Over the last 2 seasons the 2 consistent posters in midfield of shots on target have been Alexis Sanchez and Mane.
Sanchez still has some questions over whether he will start GW1 after having an injury but the next week will tell. Mane should start having had a good rest over the Summer. Liverpool have a plethora of attacking midfielders but Wijnaldum playing as a CM in the latest friendly was a good sign for Mane and should mean the attacking 3 mids are Coutinho, Firmino and Mane.
That’s it. Any comments from mathematicians in particular are welcome
Follow us for updates and news on http://www.twitter.com/fantasyfoot20 @fantasyfoot20 or facebook at http://www.facebook.com/fantasyfootballgeek (copy and paste into your browser)
This website is free. If you find it useful please support us via the donate button on the site
fantasy premier league League name www.fantasyfootballgeek.co.uk Code: 68142-27302
Telegraph fantasy football League name: FantasyFootballGeek League Pin: 8000919 League passcode: 2020
Sun Dream Team: name: FantasyFootballGeek League pin: D2T5ZCFN
