Community Submissions

One Stat To Rule Them All Revisited

There have been a few community posts recently claiming that Points Per Game (PPG) is a very valuable statistic.

Sadly, PPG in reality is about as a useful as a chocolate teapot in terms of predicting future returns.

This article aims to look at a variety of player attributes to see if we can find a far better statistical model to pinpoint those who will do well as Fantasy assets.

My research started with an article last year by Rakkhi entitled “One Stat to Rule Them All” in which he also begins an inquest to the best statistic to judge a player. This focused on shots on target as well as big chances being key indicators of future points and also prompted Jonty to produce this Members table based on Rakkhi’s findings.

Below, I’ve taken the top Fantasy Premier League points scorers for the last five seasons in an attempt to cross-reference them in order to pinpoint a common statistic in each grouping. In some cases I have missed off players like Dimitri Berbatov, who have since left the Premier League or where I couldn’t find their points returns online.

This article is restricted to Members. For more details about Membership visit here.

Bøwstring The Carp Active since 2011 on FFS. Occasional poster and community article writer. Twitter: @MattKearney92 Follow them on Twitter

103 Comments Post a Comment
  1. Dino
    • Fantasy Football Scout Member
    • 14 Years
    7 years, 6 months ago

    The danger with stats is they are only relevant in context.

    The context can be the area a shot was taken from, the type of shot (header, volley, under pressure or one on one etc), the opposition, the fixture location (home or away), the game state (was a normally defensive team chasing a game and leaving themselves open to counter attack), formation changes in persons team, positional changes for a player, formation changes for opposition team, personnel changes in players team or opposition team. I am sure there are many more but this gives an idea of things that can effect the stats of players and make it difficult when comparing with other players.

    I think sometimes we can't see the wood from the trees especially when it comes to form. Form is that magic thing that allows a player to convert a much higher than normal percentage of chances or set up a much higher percentage of goals per key pass.

    If you want to use stats to predict the future you need to be very careful what you are looking at as form changes conversion rates even in the best of players and even if you can allow for all the other contextual events (which I don't think anyone can) then form can still undo all your good work.

    1. Bøwstring The Carp
      • 12 Years
      7 years, 6 months ago

      Yup, hence why I think xG is a statistical model that would be most suited to taking these different factors into account

      But context adds a much more important layer of security when dealing with future expectations, etc.

      1. Dino
        • Fantasy Football Scout Member
        • 14 Years
        7 years, 6 months ago

        I like xg models but they are far from perfect either with game state amongst other variables not taken into account afaik? I'm not sure fixture strength is adequately considered either tbh but to me it is the best we have to go on alright.

  2. daviyb
    • Fantasy Football Scout Member
    • 7 Years
    7 years, 6 months ago

    Great article first and foremost. Recent discussions on this topic touches on the quality of shooting. So, what are the boundaries of available stats? i.e. can we measure 'shot quality'? Average speed per shot for individual player, average distance ball lands away from goalkeeper's start position whilst still being on target, etc..

    1. Bøwstring The Carp
      • 12 Years
      7 years, 6 months ago

      As mentioned a few times, I think (but I'm not 100% sure) that xG models present are more contextual picture for all of the shots we are talking about.

      I'm just new to the whole xG thing so didn't want to write up too much about it not knowing what it was fully about, however I can see that it has merits that surpass even the model I'm talking about here in this article, but is along the same lines

  3. Spreadsheet
    • Fantasy Football Scout Member
    • 10 Years
    7 years, 6 months ago

    You don't often see a community article with two pages of comments! Congrats Bowstring. Good discussions!

  4. RedLightning
    • Fantasy Football Scout Member
    • Has Moderation Rights
    • 13 Years
    7 years, 6 months ago

    Players with superior PPGs in the first 6 GWs will on average also score more points per game in the next 6 than players whose PPGs for the first 6 were significantly inferior.

    Players with superior underlying stats in the first 6 GWs will on average also show better underlying stats in the next 6 than players whose underlying stats were significantly inferior.

    If a player's PPG is significantly better than his underlying stats then this may be a sign that his PPG might regress.

    If a player's underlying stats are significantly better than his PPG then this may be a sign that his PPG might improve.

    I would have more confidence in a player who has both a good PPG and also good underlying stats than in a player who only has one of these, but the longer a player maintains a high PPG the more confident I would feel about his future prospects even if his underlying stats are still not impressive - and the longer a player with good underlying stats fails to convert these into points the less confidence I would have about his future prospects.

    I would be very surprised if the best FPL managers do not look at both points and stats as well as many other factors when making their decisions.

    1. Dino
      • Fantasy Football Scout Member
      • 14 Years
      7 years, 6 months ago

      Good summary