Community

The data-led optimal FPL squad structure

Community writer ritzyd discusses Fantasy Premier League (FPL) squad structure ahead of Gameweek 1 – and argues why ‘big at the back’ may not be the optimum approach.


With lots of talk about ‘big at the back’ and 5-2-3 drafts flying around, I wanted to look at what was the optimal FPL squad structure based on previous seasons’ data.

So I set about building the best FPL squad structure using statistical analysis for the 2022/23 season.

I looked at points gained by different player positions at different price points over the previous three seasons to work out the best team structure to maximise potential points based on this season’s starting prices.

To analyse this, firstly I excluded the lowest price point in each bracket (£4.0m goalkeeper, £4.0m defender, £4.5m midfielder, £4.5m forward) assigning this as the base cost of a player. To be included players also needed 500+ mins played.

I split defenders into ‘defending’ defenders and ‘attacking’ defenders based on less than/more than 40% of their touches being in the opposition half. Midfielders/forwards also needed minimum expected goal involvement (xGI) to qualify.

Using starting prices, I then split player positions/prices into the following categories:

Goalkeepers:

  • bargain (4.5m)
  • mid-price (5m)
  • premium (5.5m+)

Defending defenders:

  • bargain (4.5m)
  • mid-price (5m, 5.5m)
  • premium (6m+)

Attacking defenders:

  • bargain (4.5m)
  • mid-price (5m, 5.5m)
  • premium (6m+)

Midfielders:

  • bargain (5m, 5.5m, 6m)
  • mid-price (6.5m, 7m, 7.5m, 8m, 8.5m)
  • premium (9m+)

Forwards:

  • bargain (5m, 5.5m, 6m)
  • mid-price (6.5m, 7m, 7.5m, 8m, 8.5m)
  • premium (9m+)

From here I then ran an analysis to find out which Opta stats most closely correlated to FPL returns. For defenders, this included things like touches in the opposition half, attempts from set plays, goals conceded and shots on target, while for attacking players it included touches in the penalty area, shots on target, touches in the final third and others.

Using these stats plus expected goals (xG), I worked out an expected points set (excluding appearance points) and from this, an expected points per million (ppm) spent over and above the base cost in each position.

The sample used data from the past three seasons and contained over 500 individual player data sets.

Here are the results:

  • Bargain GK: 90 ppm
  • Mid price GK: 59 ppm
  • Premium GK: 69 ppm
  • Bargain Def Def: 50 ppm
  • Mid price Def Def: 54 ppm
  • Premium Def Def: 60 ppm
  • Bargain Att Def: 96 ppm
  • Mid price Att Def: 83 ppm
  • Premium Att Def: 61 ppm
  • Bargain Mid: 74 ppm
  • Mid price Mid: 51 ppm
  • Premium Mid: 33 ppm
  • Bargain Fwd: 71 ppm
  • Mid price Fwd: 34 ppm
  • Premium Fwd: 23 ppm

So how do we extrapolate this data to look at building our squad? Firstly we need to understand how to use it.

As an example, a £4.5m bargain ‘keeper would be expected to get 90 ppm above the base price (£4.0m) so from previous seasons would be expected to get 45 expected points (90 ppm*0.5m (0.5m being 4.5m cost less 4m base price)) above any appearance points. A premium GK at £5.5m would be expected to get 104 expected points (69ppm*1.5m) above any appearance points.

What are the takeaways from this data?

  • Premium forwards offer the least value per million spent of any position (excluding captaincy),
  • Forwards generally offer the worst points per million spent of any categories above.
  • Defending defenders all have a reasonably close points per million output regardless of price bracket so there is little point in spending budget on a premium defending defender. For example, you may as well choose Lewis Dunk (£4.5m) over Virgil van Dijk (6.5m) and use the money elsewhere.
  • Choosing a budget goalkeeper is only marginally detrimental to your side’s output. Don’t waste money in this position.
  • Premium midfielders offer the greatest overall points potential, followed by mid-priced midfielders with premium attacking defenders close behind that. These three areas are where you should focus most of your budget.
  • Bargain forwards compete surprisingly well with mid-priced and premium forwards.
  • Mid-price attacking defenders are generally overlooked but offer some great value and FPL points potential.
  • There is very little value in attacking (MID/FWD) premiums generally; it would be interesting to try a single premium strategy and transfer in/out for captaincy but not sure I’m personally brave enough.
  • Bargain attacking defenders don’t offer much. Choose mid-price or premium attacking defenders only.
  • Bargain midfielders offer good value, especially at the higher end (£5.5m/£6.0m)
  • There is a lot of talk about big at the back but mid-price and premium midfielders should not be overlooked. Throw away those 5-2-3 drafts! 
  • When looking for attacking premiums, 2x midfielders is better than 1x midfielder and 1x forward.
  • Returns on ppm are not linear as price increases every 0.5m step.

So what does this mean for squad structure? Based on the data above (and my own elements not included in this article), the most efficient squad structure if data from previous seasons holds reasonably consistent is as follows:

  • 1x Bargain GK (£4.5m), 1x Fodder (£4.0m)
  • 3x Premium Attacking Defenders (£6.0m+/£6.0m+/£6.0m+), 1x Mid Price Attacking Defender (£5.0m-5.5m), 1x Fodder (£4.0m)
  • 1x Premium Midfielder (£9m+), 2x Mid Price Midfielder (£6.5m-£8.5m), 2x Bargain Midfielder (£5.0m-£6.0m)
  • 1x Mid Price Forward (£6.5m-£8.5m), 1x Bargain Forward (£5.0m-£6.0m), 1x Fodder (£4.5m)

That comes out at a base cost of £94.0m at the higher end of the brackets. In reality, this might look something like the following (players chosen to highlight the point rather than because I think they are a good pick!):

  • Jordan Pickford (£4.5m), fodder (£4.0m)
  • Trent Alexander-Arnold (£7.5m), Joao Cancelo (£7.0m), Reece James (£6.0m), Lucas Digne (£5.0m), fodder (£4.0m)
  • Mohamed Salah (£13.0m), James Maddison (£8.0m), Phil Foden (£8.0m), Pedro Neto (£5.5m), Andreas Pereira (£4.5m)
  • Gabriel Jesus (£8.0m), Dominic Solanke (£6.0m), fodder (£4.5m)

This comes out at £96.5m, giving you £3.5m to upgrade elsewhere (likely to get a second premium in midfield or attack, ideally upgrading one of the mid-price midfielders to a premium one) but you can play around within those brackets to find what works for your own squad and the players you want to have.

As an aside, I ran a small sample size analysis on the effect of fixture strength on the numbers above (all Premier League games from 1st Jan 2022) but not in enough detail to warrant full discussion here.

However, a few things that stood out fairly clearly were that forwards’ points tend to be least affected by fixture difficulty as their points tend to be scored fairly randomly, premiums across all positions benefit most from easier fixtures, and bargain defenders tend to be generally poor no matter the fixtures.

The data was clear enough to lead to the thinking that rotating your premiums to target easier fixture runs is a very valuable tactic while buying lower-priced assets to sit and tick along. Focus transfers where they can be most effective (premium assets).

Hopefully, you found that helpful – and good luck for the season ahead.

FPL pre-season: Trossard hits hat-trick from wing-back role 2

820 Comments Post a Comment
  1. FLAMINI LAMBORGHINI
    • 8 Years
    3 days, 14 hours ago

    A) cancelo + martinelli
    B) walker + saka

    context: going for kane+haaland and no jesus

    Seeking advice!!