Hello and welcome back to FPL 2018/19!
This is one of our build up articles for the new season. Much like our future trends series from 2017/18, these articles reflect on what we learnt from the season before and how that might impact FPL managers’ strategies for the new campaign.
Points Per Cost background
On our final podcast last year, as part of our season review, we discussed the players who had performed best in regards to their points per cost.
(we have been referring to this as “points per value” but as FPL already have a “value” metric we’ve called it “points per cost’ here to avoid confusion)
In this article, I’ll be looking at two different ‘value for money’ metrics used to measure player output and performance over the course of the season.
By ‘value for money’ (or simply ‘value’) we mean player performance in FPL proportionate to their initial price. This should always be measured using the cost for the concurrent season with the performance rather than the one that comes after; though the later cost is informed by the prior season, forcing equivalence won’t reflect the true value of a player in a true sense.
For me, one challenge with the default dataset is that it includes appearance points, which skew the data to make cheaper players seem like better value for money
In order to mitigate this, as part of our analysis we will also be taking a deep dive and looking at how valuable players were if we take the appearance points out of the equation. By assessing player impact using purely goals, assists, clean sheets and bonus points over the course of the season, we saw some very interesting outcomes.
The power of the substitute: why counting appearance points in ‘value’ is a bugbear
Points per game (PPG) has always been my favoured metric for measuring the output of players, as it reflects their performance over the course of the season and factors in the fact that they could have just made a substitutes appearance, which they should rightly be punished for: a key factor in drafting or buying a player if he’s nailed or not. However, this metric still requires a little bit of tinkering with. For instance, if we rank all players over the course of the season in terms of ppg, right at the top appears Paolo Gazzaniga. Called upon in the absence of Hugo Lloris and Michel Vorm, a barnstorming performance and clean sheet meant he got a return of 9 points. He was never given the chance to repeat this performance, and finished the season with 9.0ppg – a feat not even matched by Mo Salah!
Some may therefore say points per 90 (pp90) is a better measure, but this still has an inherent flaw which I call the power of the substitute.
If we take Bernardo Silva as our example here, this is a player who was largely absent from the starting 11 for Manchester City thanks to the proficiency of wingers like Raheem Sterling and Leroy Sane. He therefore mostly made appearances from the substitute bench when appearing in a game. And he’d often, in these appearances, sneak a cheeky goal or assist in a typical drubbing of one of Manchester City’s opponents.
Bernardo highlights the “power of the substitute” flaw very well, if we were to analyse his player data purely by looking at his pp90. He scored 6.66 pp90, higher than his namesake David Silva, who outscored him by over 57 points over the course of the season but only scored 6.26 pp90. The reality was that David Silva started a significant amount of game time for City and this is reflected in his ppg was 5.8 compared to Bernardo Silva’s 3.2. In fact, if Bernardo Silva had made an appearance for 10 minutes nine games of the season, getting no attacking returns, he would have managed to still score a 9.0 pp90 for that 1 point he got each time for turning up. This is illustrated by the case of Phil Foden who has a pp90 of 16.0 (maybe in a few years Phil, we can only wish)!
So you can see I’m not exactly a fan of including appearance points in any metric that helps us quantify ‘value’. Therefore, we are going to have to dig deeper to find the best metric for analysing a player’s overall FPL performance.
Looking at the data – with appearances
However, as it’s widely used, we will first look at the the points per cost data including appearance.
This is simply calculated of dividing the total points earned by a player in FPL against the cost that they were given at the start of the season. This can help, at a basic level, pick out possible key bargains for the coming season. This is what we found:
The initial observation from this data is that goalkeepers performed extremely well, and represent best value for money in the game. However, as much as that is very useful information, we can only sadly put one in our team so won’t help from a formation perspective.
There is, however, a prevailing ‘De Gea uber alles‘ theory argued by many, which has been somewhat challenged by Lukas Fabianski, Mat Ryan and Nick “Popey” Pope proving themselves better value when spending your 100 million. This will be an unpopular viewpoint to take – not least with Tom, a convert last season to De Gea’s camp, who would highlight that he will be the best outlet into the Manchester United defence that kept a clean sheet every other game last season.
For me, I am looking at Fabianski and Ryan in a 4.5 rotation, the former who has managed an average of 9.75 clean sheets and impressive average of 126 saves over the past four seasons suggesting that despite moving clubs he should have it within him to match his average 4.1 points per game from last season.
Premium defenders occupy the top three in Nicolas Otamendi, Cesar Azpilicueta and Ben Davies. Their ouputs seem to offset their lofty price tags. In midfield, it’s all about Mo Salah’s freakish dominance last season; Pascal Gross, a hero of our Talisman Theory piece, is the nearest competitor.
But the big finding in this first analysis is that forwards have got the sharp end of the stick. Out of the top 50 players in terms of ‘value’, there wasn’t a single forward that made the top 50.
Jordan Ayew, Jamie Vardy and Roberto Firmino were the 52nd, 53rd and 54th best value (and something that we deemed worthy of a separate article which we will also be publishing shortly). Additionally, the order just looks bizarre, with JAyew up top and Ayoze Perez (124 points, 8g 6 a last season) finding himself somehow in fourth for points per cost by this way of doing it.
However, as you probably know by now, I’m not a fan of this metric. The likes of Zanka, for instance, show up in the top 10 for just showing up each game – being cheap at 4.5 million makes him seemingly appear to be better value to own than some premium assets with appearance included.
Better data – without appearances
We therefore enhanced our model to exclude the appearance points:
The data here is superior because it’s a better representation of what players added to their team without the baseline of appearance points. I believe that this is a better measure of the ‘value’ that a player provides to their team as it shows what they contribute proportionate to their cost.
The most interesting shift in perspective was on the Forward side from doing this analysis.
Firmino, Kane and Vardy all jumped past Ayew in terms of value: Firmino shows the best value of all. A 1.0 bump in price to 9.5m for next season still seems very reasonable all things considered!
A few oddballs begin to appear as well in the forward lists in the form of rotated Everton strikers Dominic Calvert-Lewin and Oumar Niasse. The single figure score that they record, though, shows just how poor the value was amongst the Forwards last season.
Perez aside, there really wasn’t anyone that could be considered good value, sadly, as a third striker. Even our favourite Brazilian workaholic Roberto Firmino had 10 midfielders and seven defenders that were better value than him using this new ‘points per cost minus appearance’ metric to ascertain value.
In midfield, Salah’s crazy season is again highlighted as he retains top spot, and Raheem Sterling finds his way to second place. In an interesting correspondence to the Talisman data, Gross still holds his value very well; but he’s still some way off Raheem and Mo. At 9.0 and 8.0 initial evaluations for those two, and Pascal at 5.5, they were priced amazingly in terms of their points output. This underlines the power of spotting undervalued players year on year and investing in them early before their prices balloon – and we will try to do this throughout our preseason content and podcasts.
Defence is the second bottom for value overall; the same three (Otamendi, Davies and Azpi) keep their place, though interestingly Davies and Azpi swap places, displaying how Azpi relied on appearance points that bit more than Gentle Ben to achieve his points score. It’s also interesting to note that all top 10 defenders are from big clubs, showing how cheaper defenders struggle to outperform their low price tags.
In goal, Pope leapfrogs Fabianski to take top spot: again, this shows how much appearance contributed to Fab’s initial score when it was included, and showed what a great bargain Popey was at 4.5. Keeping 11 clean sheets is clearly the driving factor here. Despite a 5.5 price tag, DDG’s performance in 3rd is in keeping with the stellar season he had. Ederson’s clean sheet numbers also see him to fourth. It’s a real mixed bag for them and hard to draw conclusions on from this data in terms of what the best choice is (we’ll be releasing an article on this in due course).
And, underlining how much he dominated, Salah’s dominance was even more enhanced with a PPV minus appearances of 26.00: a whole 5.22 points higher than his closest rival Raheem Sterling.
Here’s the top 20 for points per cost last season so you can see how they all fit together:
In terms of positions, I think we can take the following conclusions:
Between the sticks, Pope and Fabianski show the value 4.5s can show between clean sheets and saves. David De Gea’s great season sees him in third. We’ll delve into this further in another article going forward.
In terms of defenders, the premium defenders rule the roost in that position. Therefore, it might be worth investing in at least two for your FPL team in game week one, focusing in on those who are both nailed as a hygiene factor but also offer that little bit extra be it in the shape of clean sheets plus goal or assist potential. This can be a highly efficient way of spreading the cost of your team.
Midfield wise, the Salah-Sterling-Gross triumvirate shows the importance of identifying underpriced players early and getting them in. We (and you, no doubt) will be scouring the data for these guys but it may be the case that as the season kicks off one or two men start to outperform their price tag. Don’t be stubborn about it if it looks to be happening – as we saw with those “Salah dodgers” last season, refusing to join the bandwagon can hurt your rank.
Whilst there is a case still for premium forwards as captain options and key members of your FPL team in terms of the likes of Kane and Firmino, the forward does seem to be becoming less crucial to FPL success (I will expand on this in a separate piece as I think this is a subject definitely worthy of a deep dive into the data). Below a band of 5, there really was not much value in the position and it remains to be seen whether including Arnautovic and Zaha here will help or hinder them and the classification as a whole.
Either way, I hope you can see the merit in removing appearance points to gauge the true value players represented relative to their cost.
The big takeaway from this analysis is that picking those valuable players is critical to FPL success. Many people were happy, for instance, to pump money onto players like Hazard and Sanchez last season but the output they actually offered didn’t justify their cost. Ultimately, in terms of providing value for money, players half the price of the premium options such as Groß and even Milivojevic offer more in terms of value.