Determining FPL Player Value: Points-Per-Million and Points-Per-90 Data
Special guest post by Nicholas Oo
In an attempt to tackle FPL from the stats point of view, I've created a Points-Per-Million and Points-Per-90 spreadsheet that I thought might interest you guys. FPL after all is a stats-driven game so I thought I'd try analyzing things from that angle. This spreadsheet aims to calculate and compare the "value" of players: penny for penny, who's actually worth it?
So now that you know what this is about, let me give some insight on how you can use this chart. In this spreadsheet, you'll see a total of 6 tabs and here is a short explanation of their usage.
1) Sorted by Team and Position
1st tab is just a compilation of stats according to team and position. No ranking/sorting done, but an easy way to find a certain player.
2) Sorted by Position and PP90 17/18 (>1710 min player)
The main tab of concern. Here, you'd be able to see players ranked firstly by their position, followed by their points per million per 90 minutes of play. Furthermore, there's a filter that removes all players that played less than 1710 minutes (which equates to half a season of games in minutes, i.e. 90min * 19 games = 1710min). I'll explain more later.
3) Sorted by Team and PPM 17/18 (Total)
This tab sorts by team, then ranks by their points per million (total), not taking into consideration their minutes played.
4) Sorted by Position and PPM 17/18 (Total)
This tab sorts by position, then ranks by their points per million (total), not taking into consideration their minutes played.
5) Sorted by Team and PP90 17/18
This tab sorts by team, then ranks by their points per million per 90 min.
6) Sorted by Position and PP90 17/18
This tab sorts by team, then ranks by their points per million per 90 min. Similar to tab 2, except with all the players instead of just those that played 1710 min.
Note: this database only takes into consideration the 17 teams that remained in the premier league. Due to access to data on the FPL site, I wasn't able to include the promoted teams of Newcastle, Brighton or Huddlesfield. I've also done my best to keep players that have transferred out of a relegated team and into a premiership team (such as Defoe). Sorry if I missed any.
Now into some actual beef. To show you how to make use of this chart I'll use two examples to illustrate.
1) Comparison between closely priced players
Many people have been touting Kevin De Bruyne almost as a "must-have" for the coming season. However, if you look at stats based on last year, even with his stellar assist record, you'll notice that he isn't the best choice. Referring to the chart below. I've pulled out the premium (>=8.5) midfielders and grouped them together just to have a closer look at their stats. This particular picture is filtered by the PP90 stats, based on their price at the start of the 17/18 season.
PP90 is a fairer comparison among players most of the time due to the differing amount of minutes that they've played. If for example, we were to take the PPM values to compare, Sanchez and Siggy both would have shot up the rankings purely because they've both played over 3200 minutes and thus have higher point totals. Looking at the stats, KDB actually is ranked 7th on this list of 11 midfielders in terms of PP90. In terms of value, KDB actually provides less than any of the 6 above him.
Of course, data is a dead and rigid, so you would have to apply your own FPL manager thought process to this. For example when comparing Sane to De Bruyne, it would seem that Sane has that slight edge on the PP90 value, but it is also important to note that Sane only played 1781 minutes last season. Though a major portion of that might be due to injury, the onus is still on the manager to determine whether a player is prone to rotation. In my opinion, KDB is way more nailed on than Sane at the moment and hence I would still choose KDB over Sane here.
2) Using PP90 values to predict possible value in youth or players that were previously substitutes
Hopefully, I haven't already lost you guys amidst all the charts, numbers and walls of text. This is actually quite an interesting point. To illustrate this point, I'll pulled out 6 players in the chart above. On first look you'll notice that all these players have INSANE PP90 numbers. Comparing them to the premium midfielders in the chart above, they're way ahead of them and in RLC's case, several times their value. So how can we use the stats of these types of players to our advantage?
Let's first look at the 2 defenders. Both Ake and Trippier both registered under 1000 minutes last season. However, in those 1000 minutes, they've posted very high PP90 numbers. This is where external research has to be done. So for most FPL-crazy managers, we know that Ake is back at Bournemouth and nailed on. Likewise, due to Walker leaving Spurs, Trippier is now 1st choice at RB for the league's best defense. So in this situation, the stats actually help us a lot. They've identified the potential of these players to earn points when they actually do play. And since from our external knowledge, we know that they'll play major starting roles for their teams, they're definitely assets we want to look at.
Next, the 2 forwards. Both posted a PP90 of more than 1, yet their situations can't be further from each other. Rodriguez most likely has a starting spot at West Brom while Giroud is relegated to a more permanent bench spot with Lacazette coming in. This example serves more as a reminder that data could be tricky. Rodriguez got those stats in under 1000 min at Southampton. Can he replicate that as a starting striker at West Brom, with a less talented support cast behind him? Giroud will almost never see a starting role again, so despite having a high PP90, we should know to avoid him.
Lastly, the midfielders. RLC is an extremely popular 4.5m bench fodder midfielder for this season and rightly so. He's likely to feature in Palace's midfield and in an advanced position. PLUS he has a insane 3.429 PP90 stat. But with just 35 minutes played, that stat is way overblown and should totally be ignored. Nonetheless, I would still recommend him as one of the options for your 5th midfield spot, just not justified by the stats in the spreadsheet. Cesc is an interesting case. I brought him in as part of my explanation because I had him in one of my previous drafts. He has an insane PP90 and is a pure assist machine when he plays. However, it's all about how nailed on he is, and at which position.
I've been lucky enough to watch the Chelsea-Bayern game live and did not like what I saw. Fabregas was extremely happy sitting back as a deep-lying playmaker of sorts. But due to Bayern dominating the game, he didn't get very involved in the attack at all. In face Kante often showed more impetus than he did. That really turned me off Fabregas to start the season. So even though he's cheap at 7.0m, I'm holding off on getting him.
Alright, so I've highlighted 2 main uses for the spreadsheet that I've made. If anyone has any questions regarding this spreadsheet and how to use it, please feel free to drop me an email at email@example.com and I'd be more than happy to answer them.
Please also drop me any feedback on how I could improve this spreadsheet for future seasons or suggest interesting ways to use it. Cheers!