Community

The secret formula to crack FPL

Jay Egersdorff famously once said “FPL is a game of probability”. But what if you could use the wealth of data we have accumulated so far this season to swing the odds in your favour and storm to success in the final third of the season?

If there is one stat to rule them all, it could well be FPL Goal Involvement. If a team is going to score, who is the most likely player to be involved in the goal or the assist? A key component of ‘Talisman Theory’, this stat matures like a fine wine. The more gameweeks you give it, the more reliable it becomes as an indicator of future performance. But what if you could take the data and the stat one step further to predict what will happen between now and the end of the season? Amongst the realms and realms of stats, could we have found a hidden algorithm that reveals the secret formula to FPL success?

Which teams will score the most points between now and the end of the season?

the-secret-formula-to-crack-fpl

In the above table, we have looked at how many goals and assists each team has scored to date. Dividing this total by the number of games played, we get an average figure for how many goals and assists they get per game. Man City come top with 4.96 attacking returns per game while Crystal Palace and Watford are joint bottom with a meagre 1.65 a game. This highlights one of the big flaws of Talisman Theory. It’s all well and good Wilfred Zaha being Crystal Palace’s talisman, but if his team’s total attacking output is puny and listless, then his FPL points are likely to be few and far between.

Taking how many goals and assists a team has gotten on average per game and multiplying it by the number of games they have left this season, we can project how attacking returns they are likely to get in the next 12 gameweeks or so. And if we can layer on a player’s FPL goal involvement on top of this, then that’s when things get really interesting.

Which players will score the most points between now and the end of the season?

the-secret-formula-to-crack-fpl 1

*Rashford faces a period on the sidelines and may therefore struggle to hit the predicted levels.

Let’s stick with the Man City example.

So far they have scored 4.96 goals and assists per game. 4.96 multiplied by the 13 games they have left gives a total of 64.48 goals and assists to be had in the final third of the season.

Now if Aguero has a 53% FPL goal involvement, we can expect him to get 53% of his team’s 64.48 goals and assists. Obviously he can’t score AND assist a goal by himself, so if we divide this figure by two to assume he either scores the goal or gets the assist, then we can project 17.09 attacking returns from him between now and the season’s end.

In second place is Vardy with 13.39 attacking returns and Salah takes bronze with 13.04 attacking returns.

Obviously none of us have a crystal ball and no projection is 100% fail safe. Caveats like injuries and players being rotated and not starting every game will have to be taken into account. But it will be interesting to return to this article on May 17th and see how close these predictions actually were!

114 Comments Post a Comment
  1. harris
    • Fantasy Football Scout Member
    • 5 Years
    10 days, 11 hours ago

    Useful stat, but ignores the role of form which to my mind is the crucial ingredient in a good fpl season. Knowing when to get in and out of a player is the art, that trumps the science. Vardy.. , Salah, ..timing is everything

    1. FPL Virgin
      • Fantasy Football Scout Member
      • 3 Years
      10 days, 10 hours ago

      Fair point, thanks for reading.

  2. The gospel according to Mar…
    • Fantasy Football Scout Member
    • 3 Years
    10 days, 10 hours ago

    Fair play you've put some work into this. I tip my hat to you. Personally I take a slightly more impulsive approach to my team decisions. Maybe that's where I'm going wrong.

    1. FPL Virgin
      • Fantasy Football Scout Member
      • 3 Years
      10 days, 5 hours ago

      Thank you for your kind words, good sir!

  3. jphelps13
    • Fantasy Football Scout Member
    • 1 Year
    10 days, 7 hours ago

    Aguero has 16 goals and 3 assists, totalling 19. Man City have scored 65 goals. That's a contribution of 29%, why not just use this instead of the 53%/2?

    1. FPL Virgin
      • Fantasy Football Scout Member
      • 3 Years
      10 days, 5 hours ago

      Because he wasn't on the pitch for all 65 goals.

      1. fatty100kg
          9 days, 13 hours ago

          There is a slight ditch in the formula; it is not realistic that Aguero would play 90mins for all the reminding games even if he was fit. Aguero has come i as a sub or has been subbed so much this season.

          To get more realistic view (especially for Aguero, eg. for Vardy I wouldn't bother) you should count more realistic "during how many City goals Aguero will be on the pitch" than that 64.48. You could count that by checking how many goals/assists City got after Aguero was subbed or before he was subbed in.

          Otherwise great job! Just started thinking that Aguero as he seemed too good to be true - and was if agree with my points.

          1. FPL Virgin
            • Fantasy Football Scout Member
            • 3 Years
            9 days, 5 hours ago

            Thanks, you can do the same analysis with KDB and not Aguero who is more secure in the team, he comes out at around the 13 mark.

    2. pavlev
      • Fantasy Football Scout Member
      • 9 Years
      8 days, 14 hours ago

      Interesting. If you are looked for a direction to extend this analysis I would recommend two potential directions:
      1) adjust for the quality of opponent. For example the same table could be created by goals conceded by team , then you can adjust to get per game expected goals and add them up. For example if the average team conceded 2.6 gpg but Liverpool conceded say 1.3) and we are talking city (4.96gpg) v Liverpool , you would get city expected goals /assists at 4.96/(1.3/2.6)=2.48 for that game. A lot more faffing but that will introduce the fact that say sou and pool have better than average opponents.
      2. Introduce form. Rather than a simple average of season so far you could weight recent games , a simple way of doing this is with an update rule new_avg = (1-a)*old_avg + a*most_recent_result. Here, a is a parameter between 0 and 1 you need to set so we Introduce model complexity.also note this is inconsistent with extension 1 however.... In fact what you really would need to measure form would be a new stat per game called "goals/assists relative to expected" based on the rolling average stat of the oppo defence. Starting to get more complicated now though!!

      Interesting read though. These sort of stats would be exceptionally useful for anyone wanting to maximise their chances of doing well!

      1. FPL Virgin
        • Fantasy Football Scout Member
        • 3 Years
        7 days, 22 hours ago

        Thank you. Two great points there.