Fantasy EPL GW6 Recap and GW7 Algo Recommendations

If this is the first time you land on one of our Fantasy EPL Blogs, you might want to check out some of our original blogs in my Medium archives to get familiar with our overall approach and the improvements we’ve made to the captain/team recommender algorithm over time.

Top 100 Managers FPL Team Stats for GW6

Despite the many premium player blanks (Lukaku, Ronaldo, Bruno), the Top100 FPL Managers in the world had a decent week with average score of 66pts. Still a decent amount of top players scored in the 50–60pt range, so we’d say if you scored near 55pts and above you did well. 1 2 3

Most Selected Player Combo by Top 100 Managers

We used the data for most selected players for each position by the Top100 managers to create the hybrid team below:

Top 100 Team Formations

Looks like most top players are using very attacking 3-4-3 or 3–5–2formations. So, far this seems to be a great strategy given the amount of goals we’ve seen since the start of the season.

GW6 Performance Recap and Overall Stats

We did not have a great week, but still managed to score above AVG on two of our teams. We’re very happy the small risk we took with making Sarr captain paid off, which is what made the difference between our top team and some of our other configurations. We were also happy to see Jota and Jimenez finally score, so overall we’re happy, despite the premium blanks.

We also participate in the FanTeam version of the FPL, where we did not do that well despite making Sarr captain, because of Bruno missing the penalty and lots of blanks in defence 🙁

Useful Stats to Inform our GW7 Picks

Let’s start with the Adjusted Fixture Difficulty Rating (FDR) for the next three game-weeks below. We also added a reverse Easy Fixtures Rating for the next five gameweeks to use in combination with the FDR score.

Fixture Difficulty Rating Next 3 Games

Looks like MUN, BRE, LIV, WAT, EVE, and MCI have some tough games coming up, so we might want to be careful investing in players from these teams. CHE, WOL, LEE, NEW, BUR, AVL, and SOU seem to have easier schedules over the next three weeks, so our algorithm might favor players from those teams.

Easy Fixtures Score Next 5 Games

If you want to think more long term — 5+ GWs ahead — then it looks like it’s worth adding some ARS, EVE, and WOL players to the mix already mentioned in previous FRD graph. Inversely, you might not want to invest too much $$$ in MUN, BUR, NEW, BHA, and LIV in the next 4–5 GW.

Bookie Odds

For this GW we will try to stack up on players from teams that have a higher than 50 % chance of winning such as CHE, MUN, WOL, LEE, WHU, BUR and TOT. We should try not to have many defensive players from SOU, EVE, WAT, BRE, NOR, and AVL.

Teams with a higher probability to draw, especially when the game is combined with high odds for Under 2.5, might be good for selecting defensive players because if the game ends 0:0, that will result in lots of bonus points. Combined with the Under/Over Graph below, we can identify the following games with a higher probability of at least one clean sheet — CRY-LEI, ARS-BHA, or BUR-NOR.

Inversely, we might want to have more attacking players from games with high odds for Over 2.5 such as MUN-EVE, LEE-WAT, LIV-MCI, CHE-SOU, and WOL-NEW.

Referee Stats

From the stats below it appears that there is a higher chance for a penalty given in games: CRY-LEI, BHA-ARS, MUN-EVE, and LEE-WAT, so we recommend having penalty takers from some of these teams. Looks, like the games LEE-WAT and WOL-NEW have refs that like to give a lot of cards, so expect to lose some points from yellow/red cards.

Projected Starting Lineups

Before we run our final team selector, let’s take into account the predicted starting 11 for each team and the doubtful/injured players.

Please pay attention to the doubtful column above and the Injuries table below to try to avoid players that might miss next game.

Team Cumulative Stats

The stats below allow us to identify which teams have high Possession %, Passing %, Shots on Goal, Aerial Duels Won and Shots-to-Goal Conversion Rate. In terms of combined Rating we see LIV, CHE, MCI, MUN, WHU, BHA EVE, and BRE as the teams with the best combined player EPL starts so far.

Top 10 Teams by SpG and ShotConvRate

The teams that have taken most Shots on Goal so far are LIV, MCI, WOL, MUN, WHU, CHE, NEW, LEE, and ARS. Of those, CHE, EVE, MUN, WHU, EVE, AVL, and BHA seem to be the most clinical in front of goal with the top 3 conv_rates of 0.92, 0.89, and 0.82. While LIV, SOU, CRY, and WOL have some of the lowest conversion rates.

Top 10 Teams by Passing % and Possession %

The teams below seem to control the ball better and make less passing mistakes, which can result in a higher probability of clean sheets and goal opportunities created. Teams spotted in both charts seem to be playing with the ball a lot — MCI, CHE, LIV, MUN, BHA, LEE, WHU, LEI and WOL.

Top 10 Teams by Yellow Cards

You might want to be careful having many players from teams — WOL, AVL, LEE, SOU, EVE, NEW, and WAT as they tend to get more YCs.

Team Cumulative ROI Stats

The table below reveals which teams are considered good investments overall, and which teams have a lot of overpriced underperforming players. Teams are sorted by avg_pts_per_player, so to no surprise MCI, BRE, LIV, CHE, WHU, EVE, and MUN are the teams at the top of the list, since they have exceeded their expected performance given their player prices. Some of the more overpriced, underperforming players can be found in NEW, WAT, NOR, ARS, BUR, and LEE so it would be a good idea to be very selective with which players you pick from those teams.

Defensive vs. Offensive Team Stats

So far, having offensive players from LIV, MUN, CHE, WHU, MCI and EVE seems to be a good investment.

While having too many offensive players from NOR, WOL, ARS, BUR, TOT and SOU seems to be a poor investment.

Having defensive players from MCI, CHE, LIV, BRE, and WOL seems to be a good investment so far in the season.

While having defensive players from NOR, LEE, NEW, BUR, and ARS seems to be a poor investment.

Important Player Stats

Table below takes a combination of important game stats to rate player performance so far in the season. Below we can see the Top 10 rated players based on their combined offensive + defensive stats.

Top 25 Players by Shots Per Game

Below are players that have been aggressive with taking shots on goal so far this season, thus having higher xG — Mane, Antonio, Salah, Lukaku, Benrahma, Bruno, Sarr, Greenwood, Toney etc.

Top 25 Players by Passing % Success

Below are players that are more likely to complete many succesfull passes and gain points from assist + bonus — Rodri, Rice, Kovacic, Duffy, Grealish, Silva, Pogba, Cancelo, Alonso, Greenwood etc.

Top 25 Players by Aerial Duels Won

Below are players that are good in the air — both defensively, but also more likely to score during corners or crosses in the penalty area — Toney, Duffy, VVD, Lukaku, Cancelo, Antonio, Pogba, Traore, Mane, TAA, Alonso, Doucoure, Gallagher etc.

Captain Recommender

Our approach takes the predicted points for the upcoming game, probability that the player takes penalties, corners, or free kicks, a coefficient for the player’s aerial threat from the past 4 seasons, the likelihood of their team scoring 2 or more goals, and blends all of those in a normalized way into a final captain_choice coefficient. The coefficient is then discounted by an opponent_resistance score, based on the player’s next opponents adjusted FDR and normalized score for defensive strength this season. Example of what the Pandas DF looks like below:

Top 25 Players by xG, xA, and xPoints Next Game

Set Piece Takers and Aerial Threats

It is also good to consider set-piece takers and players with high penalty area threat coefficients as your captain, since they get a lot more opportunities for offensive points.

Based on that formula, here is the list of the Top15 recommended captains for this GW. Lots of good options there, so not an easy choice by all means. Our recommender thinks you should go for Wood, Lukaku, Antonio, Jomenez, or Ronaldo as your top 5 choices.

Predictive Models (Player Stats)

It’s time for the crown jewel of our Algorithm — the predicted player stats. After we layer in all the FDR, bookie coefficient, ref starts, projected lineups and injuries, there are two major metrics that we take into consideration when tuning our Team Optimizer for the next n-gameweeks team selection — predicted total points and expected value (ROI). Below are the stats for each metric, also broken down by position.

Projected Total Points Next 5 Games

Projected Value (ROI) Next 5 Games

Note: Typically we design our team as a combination of the two stats above in order to balance our budget between expensive players with high points expectation and less expensive players that are also predicted to generate good value for their price over the next 4–5 weeks.

As you can see there is a large number of options we can choose from for each position, so we will be plugging a lot of the stats above into an Optimization Function in Python, which will output the team with the highest expected total points, given our budget constraints and other metrics that go into our decision making process. Some of the preliminary filters, applied before the Team Selector Code kicks in, include:

  1. Exclude Injured or Suspended players
  2. Exclude Players from teams with high FDR
  3. Exclude Players with an upcoming blank week
  4. Cannot have more than 3 players from the same team
  5. Must have 15 players total (GK=2, DF=5, MD=5, ST=3)

Optimize Budget for most used formation

Most used formation by Top100 players last week was 3–4–3, so we will present an optimized team for that formation. The model first looks at parameters that tell it if it should optimize towards full squad of 15 players, or towards a formation with 11 key players and 4 cheap fillers. For the fillers, it first looks at preferred formation and uses that to decide how many fillers to get per position. The model then subtracts the total amount spent on the 4 fillers from our initial budget and spends the leftover budget on the key 11 players, given the optimization function and model constraints.

Example1: Optimize towards max expected points next 3–5 weeks

For an optimal 3–4–3 formation you can bench Foster, Livramento, and Laporte for the upcoming GW.

Example2: Optimize towards max value for all 15 players

This is pretty a pretty balanced team of 15 players with potential +60mins of play and point-returns with 0.3M left in the bank.

Our Team for GW7

We will always use our top-scoring team from last week, and try to do a maximum of 1–2 transfers. We had saved 2-free, so we can get three Chelsea players — Alonso, Kovacic, and Lukaku. For Captain we went with Lukaku, but it was a close call between him and Antonio.

Conclusion

We’re hoping our triple CHE investment pays out this week, and we had left 3M in the bank, hoping to be able to get a couple of MCI players for next Gameweek. As always, thank you for taking the time to read our blog and good luck this week!!!

Leave a Comment

Your email address will not be published. Required fields are marked *