Community Submissions

Player Value Revisited

A recent post got me thinking again about player value. In this article I want to offer a different perspective as well as consider some key concepts in Fantasy budgeting such as value over replacement player (VORP) theory.

How much is 0.5 budget worth?

If you are going to finish in the top 1,000 in the Fantasy Premier League (FPL) you need to achieve around 0.58 points per 1.0 budget each Gameweek before captaincy. Taking into account captaincy, you should then be near to the 65 points per Gameweek achieved by  last season’s winner Simon March (aka Fantasy Football Scout Member Dufflinks).

This increment of 0.5 is therefore worth around 0.29 points per Gameweek to a top-1,000-ranked FPL manager. Multiply by that 38 and that’s 11.02 points per season.

Obviously transfer value changes as the season wears on and there are other effects too, such as double Gameweeks. This year we also have chips offering further chance of inflating points such as a triple captaincy score. I think, though, that 10-12 points per season seems a fairly sound guideline for the value of the basic 0.5 unit.

Value over replacement player (VORP)

Taking midfielders, let’s say you could have a 4.5 player that will probably start every game and is not a card magnet. They can be expected to get the same appearance points as Hazard will (and they may also get the occasional goal, assist or clean sheet). We can therefore discount appearance points from our calculations of value. What we will be interested in are “earned” points – in the case of a midfielder these will be from goals, assists, clean sheets and bonus, minus cards (and any penalty misses).

A working yardstick

Putting these two concepts together we can now estimate the “earned value” required from our 0.5.

OK, so the mimimum spend on our squads is circa 67.5.  That would be 15 x 4.5 players. That squad should be enough to give you, say, 30 points per week (22 appearance points plus 8 earned points from the occasional clean sheet, goal or assist from a 4.5 player.). In actual fact you’d probably need to spend slightly more than 67.5 because starting strikers at 4.5 are few and far between, although you might get away with a couple of 4.0 players in defence.

Additionally, the eight earned points allowed for is an estimate; it is not precise. But 67.5 for 30 points will be a near enough figure to use for the purpose of estimating value unless someone wants to undertake the massive statistical trawl required to make it a little more accurate. It will serve my purposes here.

Of the 58 points per Gameweek we looked at as being needed before captain doubling, 30 should therefore normally come from baseline points (from your baseline 67.5 squad).  That leaves 28 “earned” points required from our 32.5 of discretionary budget. This means our discretionary 0.5 needs to deliver around 0.43 earned points per Gameweek (around 16-17 points per season).

So, let’s take Eden Hazard as an example and ask ourselves two questions. Firstly, is he good value at 11.5?  And second, is he good value at 11.5 compared to Cesc Fabregas at 9.0?

Starting with the first question, we know that a 4.5 midfielder should give us the appearance points+ and that we can’t pay less than 4.5 anyway. So Hazard has to justify 7.0 of discretionary spend of our budget through “earned” points.  To do that he must average at least 14 x 0.43 = 6 points per GW excluding appearance points. To do that he’d need to manage 304 FPL points in the coming season. Will he do that?  Very unlikely. Is he good value at £11.5m? No.

Second question. OK, Hazard has a price tag 2.5 higher than Fabregas.  To justify that he’d need to produce 5 x 0.43 points = 2.15 points per GW more than Fabregas, or 82 points more over the season.  Will he do that?  Probably not. Who is better value? Probably Fabregas.

Captaincy

I haven’t mentioned captaincy since the very beginning of this article. So now I must. Clearly the calculation of value is different for those players you would select as a regular captain option. To keep it very simple, we can afford two (maybe three) of those in our squad and they should be there on the basis of points, not value. For captains you want the highest point scorers, full stop. Because with captain doubling there is no way they won’t deliver value.

So, my earlier comments about Hazard not being good value at 11.5 don’t mean he’s a poor selection. He is one of a small handful of top-range captaincy contenders in the game and will very likely end up in the top five FPL points scorers. If you are going to captain him regularly he’s almost certainly worth his price. If not, he’s almost certainly not.

Conclusion

This article began in the attempt to quantify the value of our budget, so that comparing between different player options can be based on more than instinct. Hazard was only used as an example. I do believe that the ability to ask yourself “will this 9.0 player deliver 1.72 points per GW (65 points over the season) more than this 7.0 option?” gives you a different means of making judgements.

Couple it with studies of player PPG (points per game) and add in the effect of transient circumstances like good or bad fixture sequences and it certainly gives you an alternative (perhaps more neutral) way to look at squad selection in any case.

48 Comments Post a Comment
  1. J0E
    • Fantasy Football Scout Member
    • Has Moderation Rights
    • 15 Years
    9 years, 3 months ago

    thanks for submitting, interesting stuff.

    1. Limbo
      • 14 Years
      9 years, 3 months ago

      Agreed. Makes for an interesting read. This is the sort of stat stuff that I use to choose between low-key players, building a team around the bigger hitters. Thanks for the energy 🙂

  2. Rödallegabomb ᴿᴮ
    • 12 Years
    9 years, 3 months ago

    Thanks, good read.

    Played around with this stuff last season, made some interactive tables which were quite popular.

    Might concoct something similar this time around, but not much time to do so. They are helpful, but stats are stats and can't always be relied upon. Will be interesting to compare the tables best picks to the eventual best picks from last year.

    1. Limbo
      • 14 Years
      9 years, 3 months ago

      Please do. Love a proper stato adjustment to the squad 🙂

  3. Cpt Crunch Scott talent
    • 10 Years
    9 years, 3 months ago

    Love it.

  4. John t penguin
    • 10 Years
    9 years, 3 months ago

    does minus points impact on this in any way.
    read it couple of times and can't see it mentioned, would probably require the same effort to get correct figure as your estimated 8 points but at the very least it would reduced your starting 30 points by possibly 4-5 points per week

    1. Ruth_NZ
      • 9 Years
      9 years, 3 months ago

      I factored it into my estimated 8 points. But that is an estimate, it might be 6, it might be 10, I don't have the time or inclination to do the work needed to narrow it down.

      Whether it is 6 or 10 to be added to the appearance points only makes a marginal difference, however, for my main purpose which was to put a workable number on the value of the discretionary £0.5m. It is going to be roughly 0.4 points or so per GW. Having a ballpark number for that allows the rule-of-thumb comparisons I want to be able to make when considering player value.

  5. Sloopy
    • Fantasy Football Scout Member
    • 12 Years
    9 years, 3 months ago

    Very interesting! Thanks!

  6. Pompel
    • 10 Years
    9 years, 3 months ago

    Excellent analysis - you're not making the Hazard decision any easier!

  7. I am 42
    • 11 Years
    9 years, 3 months ago

    interesting

    depay vs rooney
    chaldi vs kane
    hendo/milner vs benteke

    whats your pick?

    1. I am 42
      • 11 Years
      9 years, 3 months ago

      for me

      rooney
      chaldi
      benteke

    2. FPLord
      • 14 Years
      9 years, 3 months ago

      strikers all the way but that answer won't help you budgetwise...

  8. tm245
    • 13 Years
    9 years, 3 months ago

    I really like this approach. I would also suggest that value might be calculated in a different way:

    Instead of filling in every slot with a minimum price player, maybe fill in the bench spots at those minimum price points and then calculate the starting spots based on your remaining money.

    X/4.0
    X/X/X/4.0/4.0
    X/X/X/X/4.5
    X/X/X

    Instead of 15*4.5m=67.5 and then sprinkling in the remaining 32.5m across the squad, you would have spent the bare minimum on the bench spots and then figured out how to disburse the remaining 83.5m for your 11 spots.

    A few reasons I suggest this:
    -- a true baseline replacement for Eden Hazard might not a 4.5m player, but rather the lowest priced player an FPL manager would consider replacing him with. Basically, if I am getting rid of Hazard, who can I get away with as his replacement? Ki, MVG, Sinclair? Another flair cheap pick? Sickening but possible, whereas Bentaleb is not even worth considering as he is another bench player and you need starters. Using your yardstick would be great here because you could calculate a minimum PPG threshold worth considering for players in different positions, likely split between defense and attack.
    -- this also pushes back against the value calculations for defenders people have posted, because those bench spots are sunk costs and must be considered when determining defender value:
    the question might not be Ivan vs a 4.5 defender, or Ivan vs a seventh attacker if you are going for a different shape than 3-4-3, but
    Ivan and his bench fodder vs two 4.5m rotating defenders or
    Ivan and his bench fodder and your attacker(s) on the bench vs a premium attacker and his 4.5m bench fodder.

    For example, Ivan and his bench fodder cost 11.0 or 11.5 and get you 4.7ppg, whereas two rotating defenders cost 9.0m and get you around 3.75ppg if things go well. That 2.5m is the difference between Hazard and Fabregas, as you point out, or Walcott and Bolasie, etc.

    I really like the yardstick approach of using the top 1k goal, though. Hadn't thought of using a points target like that to affix some of the numbers and operate within that framework. Thanks for posting.

    1. SGTMacaroni
      • Fantasy Football Scout Member
      • 11 Years
      9 years, 3 months ago

      Interesting thoughts, i like your way of thinking and im going to try it with

      X/4.5
      4.5/4.5/4.5/4.5/4.5
      5.5/5.5/5.5/5.5/4.5
      5.5/5.5/4.5

      as the baseline to see what that might drum up

  9. FrankieTheGent
    • Fantasy Football Scout Member
    • 10 Years
    9 years, 3 months ago

    This can also be used to determine the value of having non -playing bench fodder.
    So what if you have a dead player. The extra 0.5mil in a guy able to play home and away might make that OK.

    What about injuries?
    Well the auto-sub points you get might not offset the benefit of what the injured player over achieved.
    Same for dropping 4 points for a sub cos you have bench fodder.

    What's not factored in is the price growth of these premium players vs having a rounded team.

    Right now my cheapest defender is 5.0 and my cheapest mid us mahrez so I'm really afraid if dead players but has anyone looked into whether you can do well with a hardcore 11?

  10. Numb
    • 9 Years
    9 years, 3 months ago

    Brilliant analysis. I think there is another factor that your method doesn't quite take into account though - and its difficult to describe....you also need to get high "absolute points" regardless of captaincy i.e. its not JUST about value - if you go for the best value players all across the board, you might end up with, say, 5.0 spare budget - might as well spend that on a Hazard or an Aguero

    1. Ruth_NZ
      • 9 Years
      9 years, 3 months ago

      Yes. The thing is I did this research for pragmatic purposes, not academic ones. I am not in search of the best value players as such. I am in search of the best overall value for my discretionary spend of £32.5m.

      Having a number I can use (even if a ballpark one) for the value of an £0.5 unit helps me to do that, especially when comparing one player (or combination of players) with another.

  11. SGTMacaroni
    • Fantasy Football Scout Member
    • 11 Years
    9 years, 3 months ago

    Interesting article, well done.

    Quantifying it down to 0.58 per million for top 1k is really interesting, going to do some of my own research on this tonight, my main game this year will be the sky one so im not averse to trying something unusual with fpl

  12. A-VB
    • 10 Years
    9 years, 3 months ago

    Taking cue from your article itself, I think Rooney is much better value than Hazard, with captaincy considered.

    1. Ruth_NZ
      • 9 Years
      9 years, 3 months ago

      So do I. But we should be clear that you and I are factoring our judgement into that in terms of the potential points we see both will score. Someone else may judge it differently.

      I am not really interested in applying the c. 0.4 points per GW number to last seasons points to discover which players are better value. That would only tell me who would have been better value last season. So this 'formula' does not take away the need to make a judgement call about player performance that is yet to come. But what it does do is to give a means of relating the performance you anticipate to the price you need to pay for it.

      1. A-VB
        • 10 Years
        9 years, 3 months ago

        Yea I do understand what you are saying.

        My thinking that Rooney will be better value than Hazard this season (even if he was priced at 11.5 like Haz) is based on the assumption that the former will be playing the entire season either as the lone forward or in the hole. Which I think he will, considering United's current squad (even if they sign a first team striker before the transfer window deadline) and that will also make Rooney a totally viable week in week out captaincy candidate, like Hazard or Aguero. That's the reason I said I am factoring in captaincy as well.

        Cheers!

      2. Dino
        • Fantasy Football Scout Member
        • 15 Years
        9 years, 3 months ago

        What's your thoughts on the value of Aguero then?

        Cracking article btw, I think most of the best fpl managers subconsciously do this all the time as value is really what the game is all about even if austin and kane pretty much broke the game last season

        1. Ruth_NZ
          • 9 Years
          9 years, 3 months ago

          Aguero will be the same as Hazard - poor value unless regularly used as a captain. But seeing he's probably the best captain option in the game in at least half of City's fixtures then I guess anyone that has him will captain him a fair bit.

          He'll be in my team from GW4 when I will captain him. I won't bring him in before that because I wouldn't captain him even if I had him in GWs 1-3.

  13. haginio
    • 12 Years
    9 years, 3 months ago

    Cool article with made me think of some changes.

    A: Clyne, Skrtel as solid def (besides azpi)
    B: Cedric keeping 1M in the bank.

    With Southamptons fixturelist I dont know if they are worth the extra 1M, and in any case which to choose?

    1. haginio
      • 12 Years
      9 years, 3 months ago

      With A I mean Clyne OR Skrtel of cource

  14. C_G
    • 9 Years
    9 years, 3 months ago

    ahhh finally, thoughts on this team lads?

    good to go? rate out of 10?

    Pants / Hennesey

    Azpi / Koscielny / Cedric / Huth / Francis

    Sterling / Fabregas / Walcott / Mane / Bentaleb

    Benteke / Sakho / Rooney

    Cheers!

    1. fishnil
      • 10 Years
      9 years, 3 months ago

      Very nice team. Do you think Hennesey will play?

  15. Tiamani Warrior
    • 11 Years
    9 years, 3 months ago

    I'm too lazy to actually make a team and work out the figures but I suspect that going by this logic you would have a team full of 5/6/7s value wise with about 10mil left over which is where you squeeze in as many of the hazards/silvas as possible to give you the edge?

    1. Ruth_NZ
      • 9 Years
      9 years, 3 months ago

      Probably it would be something like that. Not scientific but I put up a kind of sample the other day:

      http://www.fantasyfootballscout.co.uk/2015/07/24/moving-target-carl-jenkinson/?hc_page=5#hc_comment_10591347

      The value principle might also cause you to review whether 3-4-3 is an ideal structure.

  16. TFP
    • Fantasy Football Scout Member
    • 14 Years
    9 years, 3 months ago

    I think this is quite a useful way to think about things.
    .

  17. No Luck
    • 13 Years
    9 years, 3 months ago

    Lovely work! I really appreciate your way of investigating value like this. This and other articles have even inspired me to get the Excel spreadsheets out. The way things are panning out this season with defenders priced so cheaply, I'm seriously considering 5-2-3 for the long term.

    1. Ruth_NZ
      • 9 Years
      9 years, 3 months ago

      For practical reasons I think 4-4-2 or 4-3-3 is better. But it's all a matter of opinion.

      1. No Luck
        • 13 Years
        9 years, 3 months ago

        Yeah definitely for flexibility, but if the right £4.5m mids appear then I'm considering it.

      2. La Décima Chaser
        • Fantasy Football Scout Member
        • 10 Years
        2 years, 4 months ago

        Have come back and looked at this. You were 7 years ahead of your time Ruth. Still a superb article.

      • 11 Years
      9 years, 3 months ago

      This sounds to me like you need to factor in how rotations effect defensive value.

  18. ollie -
    • 9 Years
    9 years, 3 months ago

    Impressive. Will need to create another draft on this principal and weigh the two up.

    Is there any chance of gathering your thoughts regarding Team Value increases and decreases? It's interesting that we might start the season with non-playing 4.0 defenders/keepers and very few are confident of how the price falls work this season.

    There is a certain fear of owning Wanyama or Baker due to high ownership, but what from what I've pieced together if it works on % of players selling, this should be more favourable, correct?

    1. Ruth_NZ
      • 9 Years
      9 years, 3 months ago

      I don't know is the short answer.

      I want a 4.5 bench midfielder to meet 3 criteria:

      1. A certain starter for their club (or as good as)
      2. A good recent injury record
      3. The outside chance of a goal or assist

      If they fulfil all those I'll likely keep them all season so a fall to 4.4 wouldn't bother me really.

  19. deporhostia
    • 11 Years
    9 years, 3 months ago

    Just a slight thing to add concerning value vs absolute points/captaincy.

    I'm guessing similar research on expensive heavy-hitters would also show them to be 'poor' value, as Hazard is shown to be in the article, without captaincy.

    So if you decide to go with 3 potential captaincy options, or if you fill your team with best-value players until you have a lot of budget left to splurge on big hitters, you are going to have at least 3 heavy-hitter players who don't offer great value, but will only be captain approx. one-third of the time. (And to be honest, if you had Aguero and Hazard, the 3rd player wouldn't get much of a look in with captaincy.)

    So to get best value, I guess you would be discarding this 3rd heavy hitter (who is poor value with little captaincy potential) and then slightly upgrading your 'best-value players' into something more like 'good-value players with higher points potential'?

    Hope this makes sense, first-time poster.

      • 11 Years
      9 years, 3 months ago

      This is pretty much the pragmatic way of going about it and really I think Captaincy issue should have been raised once the yardstick was established. This is because double points means half the yardstick for captains. This means the captain only needs to earn 0.215 points per extra 0.5m spent to represent value as opposed to the 0.43 points per extra 0.5m spent.

  20. rakkhi
    • Fantasy Football Scout Member
    • 14 Years
    9 years, 3 months ago

    The biggest problem with this is time horizon. You can't say Hazard won't make 300 points for the season so I won't buy him, it is more like for the next 6 weeks or right now will having Hazard give me the highest possible? Overall it is still a constrained optimisation problem (highest points for $) but taking average points for season as a way to evaluate the pics difference between players does not give you the right answer

      • 11 Years
      9 years, 3 months ago

      +1

      I do analyses similar to what Ruth is describing based loosely off RMT numbers but with my own fixture difficulty adjustments and have my own way of simulating the time horizon for a players' output.

      Firstly I cut my maximum timeframe to the week of my next wildcard (or to the end of the season if I have none left). Currently this is aimed at the gw18 Man U/Man City/Everton/Spurs/Arsenal fixture difficulty flips.

      Secondly I multiply each gameweek projection by 0.91^n where n is number of gameweeks away it is, to dilute the value of gameweeks further down the line. From this I then have an immediacy adjusted projected score from which to compare players. If a substitution looks good in this score and for this week then I will make it, if it looks optimal with this score but not in the immediate week I will roll over my FT for a week, if it looks good but not this week and I have 2FTs I will make a single transfer that does not jeopardise this transfer down the line and roll over the other FT.

      0.91^n is also (10/11)^n which I arrived at as the chance a player would still be in my team after n weeks. This works on the assumptions that I only transfer starting XI players and that I will not make transfers for a hit.

  21. rakkhi
    • Fantasy Football Scout Member
    • 14 Years
    9 years, 3 months ago

    Sounds like an interesting approach. Have you back tested it with past data to see if your model would have given you the best transfers? Not that hard to do if you have the patience with the opta data from this site.

    My approach is more stats based building a picture of the current best performing players on shots/90, shots in box, accuracy, conversation rate and chances create and trying to fit the leading players based on that. Worked well for last year .... Only problem is doesn't work at the start of the season.

      • 11 Years
      9 years, 3 months ago

      I played it out strictly last year on a test account to rank 3745. RMT basically tries to quantify those stats you mentioned into a goals/assists/bonus projection and has a yellow/red card likelihood for each player based on history.

      • 11 Years
      9 years, 3 months ago

      Although I've not backtested to previous years, I just went with my gut on the model.

      1. rakkhi
        • Fantasy Football Scout Member
        • 14 Years
        9 years, 3 months ago

        Fair enough that is a good result if you followed the model strictly. I have commented before on why I don't like RMT, just have not found its predictive powers to be that good. I wonder if the professor does back testing and tunes the model. Maybe it even uses too much data as getting the weightings right can be hard. I'm sticking with the simple formula of got to shoot on target to score and create chances to assist. Good luck for the season.

          • 11 Years
          9 years, 3 months ago

          O_O I've just seen your history, congratulations man that's brilliant work.

          Of course there are obvious flaws with RMT, but I do believe they would crack 5K most years, its time adjustment on players moving into different roles has an obvious lag as it waits to pick up data and a few other off the pitch factors. Rooney was a prime example when moving to CM from CF and took ages for RMT to catch up with his declining baseline stats. For this reason and a few others I won't be using it strictly in my own game but it is still a very good tool for guidance.

  22. Fac51
    • 9 Years
    9 years, 3 months ago

    Hows this team look...?

    Cech / Schmeichel

    Azpil / Darmian / Kolorov / (Huth) / (Targett)

    Fabregas / Sterling / Ozil / Mahrez / (Schneiderlin)

    Rooney / Costa / Wilson (Bournemouth)

  23. La Décima Chaser
    • Fantasy Football Scout Member
    • 10 Years
    8 years, 3 months ago

    I have just gone back and looked through this again with a mind to this year. This article really is excellent. I like numbers and this takes away my biased view as I can see what the numbers are saying. I had better make sure I captain Aguero every week if I have him at 13m!! Otherwise the points haul expected of him would be ridiculous! Superb stuff even a year on!