Fantasy Premier League — Hackable?

We all love Fantasy Premier League. Atleast a lot of football fans(soccer for our American friends) do. The excitement that comes in with each new season, the possibilities of better prospects for your team, the new incoming players that could add a spark…. many things to look forward to, sometimes just a clean slate too if the previous season hadn’t gone as planned.

For most FPL fans, bragging rights are key. Office games get more entertaining as the employee gains an upper hand over the manager (for once), the incoming jokes or harmless insults are all part of the fun. Let the mind games begin, friends who would love to beat each other. Or as always, some who just give up too easily. It’s all quite enticing to look forward to. Once again, football fans have arrived at the start of a season with high hopes. The 2020/21 Season looks quite exciting with teams like Chelsea and Everton splashing left and right. Newly promoted teams like Leeds also look like an entertaining prospect to watch. Ofcourse, you can never go wrong with the strongest teams in Liverpool and City. With all this happening, it can get quite mind-boggling on how to setup your team.

Recently, I came across an article about how current World Chess Champion Magnus Carlsen rose to the top of the FPL Charts during the past season. He was leading at one point in December 2019 and finished a solid 11th place by the end of the season. That got me thinking, if someone as smart as Carlsen could have climbed up the rankings, how could he have done it? Clearly, one would have to be incredibly informative about the game or a genius on how to approach FPL or just plain lucky. Carlsen being used to master strategies leveraged a bunny out of the hat.

This all seemed interesting but coming back to reality, I’m no genius or chess champion. But what I do have is the power of data. Sure, we could call this a Moneyball attempt. What if the power of data could result in similar outcomes. So this is my experiment on attempting to play FPL for the 2020/21 Season using data and analytics. So follow along as I routinely post my strategies on how I approached each week of the season.

So how would one approach this with data? There are too many parameters to consider that would take ages to compile or analyze. Thankfully, a lot of data can be found online these days. Past matches, player ratings, performances and much more are all put up online. As of now, most of my data will come from FBref. FBref provides so much information about each player and team that you really could spend hours on there trying to understand it all.

I’ll also be using R to analyze all the data that I compile.

2020/21 Premier League Season

Looking at FPL, the first thing to take care of was to understand how we could leverage all the additional chips such as Free Hit, Wildcards, Bench Boost and Triple Captain. Another thing that’s also common in Premier League is a Double Gameweek.

A Double Gameweek is when teams play multiple fixtures during a single Gameweek due to schedule conflicts with Cup games. These weeks are crucial to scoring points and the team you have then does matter a lot. While we can predict a lot of things with data, we are still not sure how this season will pan out and until we progress a bit more forward, we have to make a few assumptions to help with decision making.

One assumption I’ll be making is to decide how long into the future to plan. Wildcards can be applied twice during the season. Once before the end of the year and once before the end of the season. Since a wildcard allows for a complete revamp of our team, we don’t need to plan for the ultimate lineup that will last throughout the season — just the first couple of Gameweeks.

Another assumption I’ll be making is form. Traditionally the season can be broken into 3 parts.

Phase 1 (Sep-Nov) — The Early part of the season where teams gel together and start to get into form. Unexpected things happen here.

Phase 2 (Dec-Mar) — The Middle of the season. This is when teams have gelled together well. The players have been playing regularly and are in their peak physical condition.

Phase 3 (Apr-May) — The end of the season. The injuries have started to pile up. Unexpected things happen here as well.

Now granted, some teams could break into form during Phase 1 or 3 but I’m making the assumption that in general, they will be more in form during Phase 2. So as of now, we could leave the implementation of the Chips to the middle of the season. There are 38 Gameweeks, if we plan to break down the season into significant areas where we may apply chips, we can assume that we will possibly apply the first Wildcard or chip after 7–8 Gameweeks. Using this assumption, I’ve compiled the schedule for the first 6 Gameweeks for each team.

The aim was to identify the potential strength for each team and how they fared when facing other teams over the 6 Gameweeks. I used the following criteria to accumulate the strength for each team.

Strength of Team = (Final standing of 19/20 Premier League) - CPL + Game Value+ Transfers + Bonus

CPL (Core Player Loss) — This is when a team had to lose a key player that helped them achieve their final standing from the 19/20 league.

Game Value — We face a complication with 4 teams not playing in the Gameweek 1 (Aston Villa, Burnley, Manchster United and Manchester City). Points need to be deducted for teams that allow for fewer opportunites to score more.

Transfers — Several teams have added talent that may not feature in Gameweek 1 but may over the 6 Gameweeks we analyze.

Bonus — An additional complication I faced is the addition of the 3 newly promoted teams from the championship. Since we don’t have data to compare them to the rest of the Premier League, we faced an issue in the comparison. Thus we have added bonus points for them to emulate their potential standing compared to the rest of the league. This is an assumption that will be removed as the season progresses with results from these teams.

Based on these parameters, we get the following result.

Key notes:

  1. West Ham United seem to have the worst possible start to the season with the strength of their combined opponents amassing the most points (NEW,ARS,WOL,LEI,TOT,MCI). Good luck to the Hammers. This is a team we may want to avoid but let’s continue to explore.
  2. Leicester and Wolves seem to have the easiest schedule during this time. Wolves only face one tough opponent in City whereas Leicester will face City and Arsenal.

While this information is valuable, let’s look further. Now let’s take a look at the what happens when we compare the strength of each team vs. the combined strength of their opponents.

Strength of Team vs Opponents

From this new result we do see similar results but if we look closely, we can gain some additional nuggets as well.

The ideal scenario would be to pick from the teams that provide the most amount of action but also the greatest probability to win. From the graph on the left, we can see that teams of strength higher that a value of 12–15 with a low fixture difficulty of less that 75 would be ideal for us.

Using that logic, we can focus on these following teams.

These teams provide the most potential for a win along with solid performances. Notice that in addition to Leicester and Wolves, a solid bet would be include players from Sheffield United, Tottenham, Chelsea or even Liverpool in your FPL Team.

A word of caution with Chelsea as they have brought in a lot of new players and this may go against or in their favor depending on the chemistry of the new players and the old.

In general, this is valuable info as we start to choose which players to field our FPL Team for the first couple of Gameweeks until we are ready to make changes.

With that said, I’ll be working on tinkering the squad as I’m sure you all will be. Follow along next week to see who I’ve chosen and how they’ve done (Hopefully well). Also included will be the strategies that went into choosing each player among others.

Good luck to all of you.

--

--

--

Fantasy football using Data Analytics to leverage points and places.

Recommended from Medium

Jamaica T10 2022, Match 8: MIT vs SRO Dream11 Prediction, Fantasy Cricket Tips, Playing 11, Pitch…

NFL Playoffs Best Ball Advance Rates

Hard work pays off for Stackhouse, Raiders

Ashton Agar strikes thrice to hand Australia an easy T20 final win against New Zealand

PAK vs WI Dream11 Prediction, Fantasy Cricket Tips, Playing 11, Pitch Report and Injury Update for…

Think you understand pain & injury? The healthiest NBA team ever says ‘think again’

Northeast Philadelphia coach pays it forward through soccer

Decoding the uniquely lonely and complicated life of Tennis through Andre Agassi’s Open

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Tom Thomas

Tom Thomas

I geek on stats • Certified IBM Data Scientist • Industrial Engineering Background • Exploring the power of data • tomthomas.github.io

More from Medium

Goodbye Ole — Always A Le

Was Bozidar Bandovic needed to be sacked? Chennaiyin FC 2021–2022 A Case Study - Part Two

Footballs underappreciated European Golden Boot winner - Luca Toni

3 things you might’ve missed from #BarçaSevilla