The approach
I decided to look at player value and try to do some analysis on using the £100m in the most efficient way possible.
Now when we look at value, the players with the best points per million are almost always the budget picks, however a team full of the best value players won’t cost very much and won’t net you many points.
The approach I decided to take was to start with a base team of 11 of the highest scoring 4.0m goalkeepers, 4.0m defenders, 4.5m midfields and 4.5m forwards. I then looked for the best upgrade to that team that offered the best value (points per million). I continued with this approach until the team value was at £100m (presuming the cheapest possible bench) or until there were no more upgrades available without breaching that budget.
This, in theory, should give you the best value 11 players while using the whole budget available.
I created a simple script to run this process and automated it for all possible formations using the starting price for the players and their final points tally for the season.
Limitations of the approach
This approach effectively gives you the best ‘ghost ship’ team not accounting for autosubs.
It presumes the best way to play is to maximize your first 11 with as little spent as possible on the bench, which may or may not be true.
Captaincy is not factored into the approach, although the highest scoring player of the season does tend to make their way into the most efficient use of funds as we will see.
It also doesn’t cater for players who don’t play consistently throughout the season (through injury, rotation or simply being a January signing).
My aim with the approach was not to help identify the specific players to select, but to identify any themes in efficient use of funds giving a higher chance of selecting the right types and combinations of players during the season.
The results 2018-19
I initially ran the formula on data from the 2018/19 season with no team restrictions and the formation that came out highest was 4-5-1 with a total points tally of 2283 excluding captain:
GK: Alisson | 5.5m | 176
DEF: Andrew Robertson | 6m | 213
DEF: Virgil van Dijk | 6m | 208
DEF: Aymeric Laporte | 5.5m | 177
DEF: Trent Alexander-Arnold | 5m | 185
MID: Mohamed Salah | 13m | 259
MID: Sadio Mane | 9.5m | 231
MID: Raheem Sterling | 11m | 234
MID: Eden Hazard | 10.5m | 238
MID: Ryan Fraser | 5.5m | 181
FWD: Raul Jimenez | 5.5m | 181
Not far behind was 3-5-2 with 2274 with Wilson up front instead of Laporte in defence.
Obviously, those results are only of limited use due to the volume of Liverpool players. It does, however, highlight the dominance of Liverpool players in FPL value-for-money last season.
Running the process again but limiting it to three players from Liverpool turned out 3-5-2 with the best score of 2196:
GK: Ederson | 5.5m | 169
DEF: Andrew Robertson | 6m | 213
DEF Aymeric Laporte | 5.5m | 177
DEF: Trent Alexander-Arnold | 5m | 185
MID: Sadio Mane | 9.5m | 231
MID: Raheem Sterling | 11m | 234
MID: Eden Hazard | 10.5m | 238
MID: Ryan Fraser | 5.5m | 181
MID: Gylfi Sigurdsson | 7.5m | 182
FWD: Pierre-Emerick Aubameyang | 11m | 205
FWD: Raul Jiminez | 5.5m | 181
Although Robertson, Alexander-Arnold and Mane came out as the optimal Liverpool players, there was very little in it. Swapping in Salah for Mane, having both Mane and Salah and dropping a defender, or even going for three at the back. All came within a few 10 points of each other.
Looking back to 2016-18
Rerunning the same process for 2017-18 gives a best result of 2136 for a 3-5-2 formation of:
GK: David De Gea | 5.5m | 172
DEF: Nicolas Otamendi | 5.5m | 156
DEF: Cesar Azpilicueta | 6.5m | 175
DEF: Marcos Alonso | 7m | 165
MID: Raheem Sterling | 8m | 229
MID: Mohamed Salah | 9m | 303
MID: Pascal Groß | 5.5m | 164
MID: Christian Eriksen | 9.5m | 199
MID: Kevin De Bruyne | 10m | 209
FWD: Roberto Firmino | 8.5m | 181
FWD: Jamie Vardy | 8.5m | 183
Going back further to the 2016-17 season gives us 4-5-1 with a score of 2095:
GK: Tom Heaton | 5m | 149
DEF: Marcos Alonso | 7m | 177
DEF: Cesar Azpilicueta | 6.5m | 170
DEF: Charlie Daniels | 5m | 134
DEF: Gary Cahill | 6.5m | 178
MID: Josh King | 6m | 178
MID: Alexis Sanchez | 11.5m | 264
MID: Christina Eriksen | 9m | 218
MID: Deli Alli | 9m | 225
MID: Gylfi Sigurdsson | 7.5m | 181
FWD: Romelu Lukaku | 10m | 221
Themes
The actual players that come out each year are perhaps not very surprising in retrospect – they are the players that did well that season. There are, however, some themes that emerge that are quite interesting to note.
Premium at the back
Every single result comes out with a premium defence. Only once do we see a defender at 5m and once a goalkeeper at 5m, all other iterations have 5.5m+ defenders as the most efficient use of funds.
Even with his monster haul of 157 for 4.5m in 2017/18, it was still worth upgrading Fabianski to De Gea for that season. That extra 1m wouldn’t have got you a better points boost spending it anywhere else on the field.
Similarly, Doherty and his 144 for 4.5m last season were not enough to get him near the best use of funds.
Overall, 0.5m spent upgrading at the back almost always nets you more points than spending it in midfield or attack. In the last three seasons, the best approach would have been to get the highest scoring defenders, no matter the cost.
Five at the back
Despite the value being in upgrading the defence, there are generally not enough high scoring premium defenders to make five at the back worthwhile. Last year 5-4-1 came close to competitive as a formation, but only if you were allowed four Liverpool defenders plus Salah and Mane (which of course, you weren’t).
Theoretically, if we had three teams with lots of clean sheets, all with attacking defenders that were fairly rotation-proof, it might work, but the last three seasons that hasn’t happened.
Formations
Formations tend to perform fairly consistently season-to-season, certainly for the last three anyway. To analyse the overall performance of each formation, I averaged the points difference for each formation when compared to the optimal formation each season:
3-5-2: 5 points
4-5-1: 14 points
4-4-2: 19 points
3-4-3: 31 points
5-4-1: 44 points
4-3-3: 51 points
5-3-2: 57 points
5-2-3: 105 points
So 3-5-2, being on average within 5 points of the optimum formation over the last three seasons, is the fairly clear winner here. Surprising for me is how close behind 4-5-1 and 4-4-2 are, and that they are ahead of the usually popular 3-4-3.
If we rank the formations in order of score for each season (1-8), and look at the standard deviation (distance to the average) for each, none is higher than 1.5, showing that there is very little fluctuation in which formation performs best: some formation are just always better than others.
3-4-3 actually has the highest variance of results, fluctuating between second best through to fifth best, showing that while it can sometime perform well, it hasn’t done so consistently.
Use 5-2-3 if you want to lose your mini-league, it ranks 8th out of 8 for each of the last three seasons.
Price bands
For those who base their team selection around price points, there does not seem to be an optimum setup that works every year. There are, however, some themes that run throughout all three analysed seasons.
More than three heavy hitters is a sub-optimum setup. There isn’t an instance of four heavy hitters working out for the best.
A price of 11m and above usually requires a score of ~220 points to justify the spend. In the past three seasons only Aubameyang squeezed into contention with 205 in the premium bracket, and that was only because too many Liverpool players did well so that you couldn’t have Salah and Mane with two defenders.
There is almost always at least one 5.5 – 6.5m candidate that comes good – getting that one right seems to be key to using your budget wisely. No matter what formation you roll in any season, the cheap overperforming attacker appears in every formation that season (Jimenez and Fraser last season, Gross in 2017-18 and Josh King in 2016-17).
Forwards are struggling
Forwards just don’t get much over 200 points, with 1 less point for a goal and no clean sheet points. It was a theme that came out last season, but the data backs it up. Mid-priced or cheap forwards are almost always the most efficient way to go.
Unless we can get a forward on fire this year, or the midfielders flop, its unlikely any will justify an 11m+ price tag.
Conclusion
The fact that the data bears out people’s recent experiences encourages me that there are some useful lessons to be learned from this way of analysing team setup.
Every season is different but there does seem to be some common ground, at least in recent seasons, that can inform decision our making.
My plan, pre-price releases, is to go for 3-5-2 with premium options across the entire back four and mid-to-cheap up front.
4 years, 10 months ago
Amazing!