#FPLMATCHUPS GW 5

Minutes per 2 Goal Opportunities FDR

This week I am going to do something a little different with this article. Rather than break down individual matchups with a tactical lens, I will be introducing a new Fixture Difficulty Rating (FDR) that I have been working on during the IB. This is completely unfamiliar territory for me, and I therefore could not have completed any of this without some very patient people helping me along the way. Please allow me to begin with thanking those people.

The Thanks:
I must begin by thanking @rogue_wee and the @uncertainty_pod. This fascinating new podcast planted the seed for this project in my mind, and I have subsequently DM’d Wee_Rogue at all hours to get his feedback on my model, without which I would have given up.

I’d also like to thank @FantasianFPL, @FPLMariner, and @barCOLEYna for their guidance and cross-referencing with their work. Lastly, I benefitted greatly from feedback from the entire core research team of @GFCT.

Ok, the models.

The Models:

Expected Goals = xG, Expected Goals Conceded = xGC, Big Chances = BC, Big Chances Conceded = BCC.

I created 2 models using similar data. One is an FDR that measures defenses for attacking players, the other measures attacking output for defensive matchups. I have always thought using one metric was insufficient, so I am experimenting with combining xG and BC. The hope here is for each stat to complement the other and fill in gaps inherent in each.

I learned that 1xG = approx. 3BC. So, to avoid over-emphasizing any one stat, the numeric values you will see in the charts below represent minutes per xG(C) AND 3B(C):

• Defense FDR for Attackers measures the number of minutes per 2 goal opportunities conceded. So, the greater the minutes, the more DIFFICULT the fixture because it takes more time for the opponent to concede goal-scoring opportunities.
• Attack FDR for Defenders measures the number of minutes per 2 goal opportunities created. So, the greater the minutes, the EASIER the fixture because it takes more time for the opponent to create goals-scoring opportunities.

You will notice the colored bars indicate this week’s matchup (GW5); the grey bars indicate the average matchups for the next 6 GWs.

6 GW Average:
My 6-game average projection is calculated using data from 3 different sets:

1. 2020/21 – weighted x3,
   1. 2020 post-restart – weighted x2,
   1. 2019/20 pre-lockdown – no weight.

The Examples:
To illustrate, we will take the first two teams in each graph as examples.

1. Southampton attackers:
   1. SOU will face CHE in GW5. CHE allow 1xG AND 3BCC (approximately 2 goals) every 138.78 minutes according to the model.
      1. No model is designed to replace critical thought. One must consider how creative the attacking team is, how clinical a player is in converting BC, and xG(C) Delta.
         1. In GWs 5 – 10 SOU will face defenses that average 1xGC + 3BCC every 132.51 minutes.
2. Sheffield United defenders:
   1. SHU defenders have quite an easy GW5 matchup from a defensive perspective. Their opponents, Fulham, create 2 goal opportunities every 195.3 min. So less than 1 goal per game.
   1. From GW5 to GW10, their defensive fixtures get significantly more difficult as their opponents will create 1xG and 3 BC an average of every 130.43 minutes.
   1. The Case of WBA attack:
      1. Burnley is missing in the DEF FDR for Attack. I made a call to omit WBA attacking data from model because their stats are so outrageously bad that they skew the data. For reference, Burnley has the worst attack listed with 1xG and 3BC every 196.61 minutes. WBA create 1xG and 3BC every 691.35 minutes! With a  bit of a deeper look, we notice that they have an xG Delta of +3.48, which is notable after only 4 GWs. These stats for WBA are unsustainably bad, and I will therefore wait until they regress before including them.
   1. Burnley’s outlook over the next six matches is nowhere near as good for this week. They will face attacks that create an xG and 3 BC every 108.86 minutes on average. A tough ask for any defense.

I will update my model weekly and share the charts in this article. I just want to remind everyone that I am by no means a statistician. I am sure this model is rife with mistakes, and I intend to use those mistakes to learn. With this in mind, I welcome all sorts of feedback and perspectives. I look forward to our discussions. Thank you again for taking the time to read FPL Matchups for GW5. Good luck, and may you pack the greens this GW!

Regards,

Gabriel Penaloza

Follow @FPLLens