44
Community

How much does fixture difficulty affect returns

Anticipating early Everton returns? Planning to steer well clear of anyone facing the top 6? Or just going gung ho with a form over fixtures mantra?

We all make assumptions about fixture difficulty one way or another in our team selections, but how much difference does it actually make to returns – is it massive or marginal? And does it vary depending on whether you’re a high scoring title contender or grinding out minimal goals and assists at the wrong end of the table?

It’s an area which I, and I think most people, often go ‘informed gut feel’ rather than anything more objective, so I decided to look at what the stats said for 2018/19…

Categorising historic fixture difficulties for 2018/19

Good attacking teams and good defensive teams are not always the same thing, plus with c.57% goals scored at home, and c.60% clean sheets, this needs factoring in too. The below tables are therefore based on goals scored/conceded over the season, and what I think is the best someone could have predicted in terms of fixture difficulty (so rather than use actual anomalies like Crystal Palace low scoring at home, which would have been difficulty to predict, I just applied the overall ratios).  I’ve sought to tier it in terms of fairly similar intervals hence bulking up in the centre where there’s a lot of fixtures of similar difficulty:

Fixture difficulty in relation to gaining attacking returns

1* difficulty (Over 1.9 goals per game)

2* difficulty (1.55 to 1.9 goals per game)

3* difficulty (1.25 to 1.55 goals per game)

4* difficulty (0.85 to 1.25 goals per game)

5* difficulty (0.8 or less goals per game)

Fixture difficulty in relation to gaining defensive returns

1* difficulty (1 or less goals conceded per game)

2* difficulty (1 to 1.4 goals conceded per game)

3* difficulty (1.4 to 1.8 goals conceded per game)

4* difficulty (1.8 to 2.3 goals conceded per game)

5* difficulty (Over 2.3 goals conceded per game)

[Happy to share specific fixture categorisations, but excluded on word count limits]

Looking at different tiers of teams

In exploring whether fixture difficulty affects teams in different ways, team by team is really too small a sample size, so instead, I tiered the teams from last year into 5 categories for each of attacking and defensive returns, e.g. Man City & Liverpool 5*, other top 6 4* attack etcc.

For each tier, I then calculated the mean attacking returns (goals, which is obviously proportionate to assists too) and defensive returns (clean sheets) for fixtures of each difficulty level, and compared this against the mean returns across all games.

An example to explain: the 4* defences of Chelsea, Spurs and Everton achieved a combined 43 clean sheets in their 114 games (returning a clean sheet in 37.7% of games).  In 1* difficulty defensive matches, they achieved 15 clean sheets in 24 games (returning a clean sheet in 62.5% of those games).  So in 1* difficulty games, they achieved 166% (0.625/0.377) of their average expected return.

So how did fixture difficulty affect returns? 

The below table and graphs summarise the findings:

[to be added]

Fixture difficulty in relation to gaining attacking returns (i.e. opponent defence strength)

1* difficulty 2* difficulty 3*difficulty 4* difficulty 5* difficulty

Average 139% 125% 105% 85% 46%

5* Attacking team 134% 141% 102% 76% 36%

4* Attacking team 154% 118% 100% 85% 44%

3* Attacking team 146% 104% 117% 79% 46%

2* Attacking team 120% 131% 94% 95% 71%

1* Attacking team 134% 148% 107% 88% 30%

Fixture difficulty in relation to gaining defensive returns (i.e. opponent attack strength)

1* difficulty 2* difficulty 3*difficulty 4* difficulty 5* difficulty

Average 173% 108% 87% 45% 9%

5* Defensive team 127% 93% 93% 93% 93%

4* Defensive team 166% 84% 106% 66% 0%

3* Defensive team 170% 114% 79% 47% 0%

2* Defensive team 174% 113% 90% 34% 0%

1* Defensive Team 234% 122% 69% 0% 0%

At the most basic level, as you can see, fixtures matter. And they matter to everyone.

Attacking returns

• Fixture difficulty seems to affect most sides very similarly.

• 4* attacking sides (top 6 minus Man City/Liverpool) go beautifully on trend as you would expect.

• For 5* attacking sides Liverpool and Man City, there is little difference in attacking returns between 1* or 2* fixtures. In fact on average the 2* rated fixtures were slightly more generous for returns.  Likewise for weaker 1*/2* attacking sides.  It is difficult to conclude categorically, and may be an outlier, but an interesting trend.  This is also played out for 3* attacking sides (Bournemouth, Everton et al) who struggled more for returns in 2* difficulty fixtures than the corresponding 3* ones.  Again, this could just be an outlier, or it could be due to the way opponents set up when they just slightly worse than the side in question set up.

Defensive returns

• With regards to defensive returns, fixture difficulty affects top teams rather minimally, but very significantly for poorer defensive sides.

• There is one outlier to the trend which is 3* defensive sides getting more clean sheets in 3* fixture difficulty games than 2* difficulty games, but I think that is due to the high level of unpredictability against sides in these categories so you would expect it to even out over a larger sample size.

What conclusions can we draw for planning going forward?

If you’re not into hard numbers (you may be in the wrong article!), the general principles to take away are:

• Over any given period of e.g. 5 fixtures, there is unlikely to be a huge swing in returns. A rubbish team doesn’t suddenly become great, and vice versa – don’t let it cloud your judgement against picking players who are good picks regardless of fixture.

• I think the best way to think about fixtures is a multiplier against what you’d expect a given player in a normal run of fixtures.  As a broad rule of thumb, excluding appearance points, think along the lines of:

  • Amazing set of fixtures: 120% returns multiplier
  • Good set of fixtures: 110% returns multiplier
  • Average set of fixtures: 100% returns multiplier
  • Poor set of fixtures: 90% returns multiplier
  • Awful set of fixtures: 80% returns multiplier

• In looking at individual fixtures, clearly things can vary slightly more than that, but the biggest thing if you’re playing the fixtures is to avoid fixtures against Man City and Liverpool, which will almost certainly result in 0 defensive returns and minimal chance of attacking returns.  Fixtures against other teams (e.g. other top 6, with the exception of maybe Chelsea at home defensively) are less extreme, and shouldn’t stop you outright from selecting a player against these teams.

• For midfielders and forwards, because fixture difficulty affects attacking returns more or less the same across teams, swapping around the highest overall scoring players based on fixtures,  particularly if it concerns captaincy, should in theory maximise transfer policy points returns (e.g. the old Kanexit rather than tinkering with 5.5m forwards based on fixtures).  Man City or Liverpool defenders in comparison are more set and forget.  Clearly more factors are at play than just this in terms of transfers, but worth considering as part of the mix.

• For defenders other than Liverpool and Man City, watch out particularly for the really easy fixture at home to teams which struggle to score, likelihood of clean sheets rocket in these fixtures (albeit sometimes from a low base).

• The figures haven’t convinced me to change my scepticism of rotation, but if you do go for it, I’d suggest the best way to utilise rotation would be 3 decent players (10th, 11th and 12th men) with 2 starters each week avoiding Man City/Liverpool fixtures.  In theory, 3 way rotation defenders focussed on maximising the very easiest fixtures could work well, since clean sheets are significantly more likely over other fixtures, but these games are few and far between (most teams have 2 in their first 10 games) so it’s difficult to sustain for any length of time.

• The fact Man City and Liverpool scored just as (actually more) highly against 2* difficulty defences as the worst defence last season means there’s a fair range of fixtures where you’d anticipate high captaincy returns (e.g. 9 points per game for Salah and Sterling) not just the really obvious ones like Norwich at home.

Suggested figures for projecting return variance by fixture

Based on last season’s figures and some smoothing out some of the outliers, some suggested projections of variance by fixture.

Fixture difficulty in relation to gaining attacking returns (i.e. opponent defence strength)

1* difficulty 2* difficulty 3*difficulty 4* difficulty 5* difficulty

5* Attacking team 135% 135% 102% 76% 36%

4* Attacking team 154% 118% 100% 85% 44%

3* Attacking team 146% 114% 114% 79% 46%

2* Attacking team 125% 125% 103% 95% 50%

1* Attacking team 140% 140% 107% 88% 30%

Fixture difficulty in relation to gaining defensive returns (i.e. opponent attack strength)

1* difficulty 2* difficulty 3*difficulty 4* difficulty 5* difficulty

5* Defensive team 127% 99% 93% 93% 93%

4* Defensive team 166% 102% 102% 66% 2%

3* Defensive team 170% 114% 84% 47% 2%

2*/1* Defensive team 174% 117% 95% 34% 2%

They could no doubt be improved, and further analysis of earlier seasons could definitely help in correcting or reinforcing the perceived outliers, but the above seems like as good a starting point as any I’ve seen.

What would that mean for returns in early fixtures?

Attacking returns vs a normal set of fixtures

Average first 3 GWs / Average first 6 GWs

ARS 82% 98%

AVL 86% 99%

BOU 103% 99%

BHA 114% 97%

BUR 105% 99%

CHE 101% 92%

CRY 91% 96%

EVE 125% 119%

LEI 91% 96%

LIV 124% 102%

MCI 93% 114%

MUN 96% 98%

NEW 100% 100%

NOR 83% 84%

SHU 118% 102%

SOU 85% 98%

TOT 99% 98%

WAT 113% 96%

WHU 91% 108%

WOL 108% 103%

Defensive returns vs a normal set of fixtures

Average first 3 GWs / Average first 6 GWs

ARS 95% 94%

AVL 89% 86%

BOU 108% 109%

BHA 129% 113%

BUR 108% 99%

CHE 90% 101%

CRY 92% 92%

EVE 102% 113%

LEI 92% 87%

LIV 97% 102%

MCI 93% 100%

MUN 114% 109%

NEW 82% 88%

NOR 97% 90%

SHU 110% 105%

SOU 89% 100%

TOT 89% 90%

WAT 129% 100%

WHU 82% 96%

WOL 113% 103%

Not actually that much difference is it? Although still notable enough to take heed of.  What a disappointingly moderate conclusion!

TLDR 

Fixtures matter, but not so much to fundamentally change good/bad picks.  Think of them as a multiplier to expected returns in general.  Avoid games against Liverpool/Man City

44 Comments Post a Comment
  1. Markus
    • Fantasy Football Scout Member
    • 10 Years
    2 months, 29 days ago

    Apologies for the awful formatting. I will see if I can upload graphs and tables elsewhere on the Web to link as they are the main bits that make sense of it all!

    1. G-Whizz
      • Fantasy Football Scout Member
      • Has Moderation Rights
      • 2 Years
      2 months, 29 days ago

      Thanks for this submission!

    2. Markus
      • Fantasy Football Scout Member
      • 10 Years
      2 months, 29 days ago

      Here are the graphs and most tables. https://ibb.co/album/nPsJ3v

      If a mod is able to add link/graphs into the article that would be amazing rather than it just saying [insert graph here]!

      1. G-Whizz
        • Fantasy Football Scout Member
        • Has Moderation Rights
        • 2 Years
        2 months, 29 days ago

        I'll see if Geoff can help but not making any promises!

      2. Geoff
        • Fantasy Football Scout Member
        • Has Moderation Rights
        • 6 Years
        2 months, 29 days ago

        Cheers 🙂

        Yeah, generally on those they have to be sent separately through email - apologies if you did I couldn't find them! Thought it would be better just to get it live, I'll get them added later tonight

        1. G-Whizz
          • Fantasy Football Scout Member
          • Has Moderation Rights
          • 2 Years
          2 months, 29 days ago

          Diamond!

          Cheers guv 🙂

        2. Markus
          • Fantasy Football Scout Member
          • 10 Years
          2 months, 28 days ago

          Thanks geoff, yeah I'd sent through word doc with graph images, but no worries. A fuller set of image files are linked if helpful.

  2. Jaws
    • Fantasy Football Scout Member
    • 5 Years
    2 months, 29 days ago

    Thx. Also the bogey team factor is very important imo. Even on a player level, Firmino for example should be captained whenever LIV plays ARS.

  3. A Fat Spanish Waiter
    • Fantasy Football Scout Member
    • 5 Years
    2 months, 29 days ago

    Love this analysis. Thank you very much. Question for others: are these ratios more or less what gets incorporated into the RMT tool on this site?

  4. A Fat Spanish Waiter
    • Fantasy Football Scout Member
    • 5 Years
    2 months, 29 days ago

    One more comment... I’m very surprised this doesn’t change your view on rotation. For the record, I have always been skeptical of rotation as well. However, these numbers seem to massively support the idea of rotation based on fixture quality (as opposed to the normal home/away rotation that seems so common for GKs and DEFs).

    1. Markus
      • Fantasy Football Scout Member
      • 10 Years
      2 months, 29 days ago

      Thanks. On rotation, theoretically it looks good until you apply to fixtures which are much less neatly organised. The best I could get over early run of fixtures was a multiplier of 140% for Watford and Brighton defences I think it was, but to get say holebas and Duffy you'd be spending as much as a Coleman + fodder who is already able to achieve above average with fixture. Doing the sums on clean sheets I'm expecting of them over the season it was more or less the same so no real advantage. The other thing, particularly for attackers is that individual player returns in a game are much more random so while you may be able to predict team returns on an individual player level you're more likely to get it wrong from a small sample size (eg game by game). See this excellent article for reference on that: https://mathematicallysafe.wordpress.com/2018/07/08/fpl-analysis-the-impact-of-fixtures-on-player-performance/

      1. Markus
        • Fantasy Football Scout Member
        • 10 Years
        2 months, 29 days ago

        You saying that though makes me rethink, I may have a play around again (and lower Newcastle defence strength!). I got burnt by rotation with gross's first monster return (20 points I think it was) couple of seasons back, although that wasn't pure fixture rotation, so ive avoided since as much for minimising emotional anguish as maximising points really!

        1. Markus
          • Fantasy Football Scout Member
          • 10 Years
          2 months, 29 days ago

          And finally, yes. Home+away poor

          1. Markus
            • Fantasy Football Scout Member
            • 10 Years
            2 months, 29 days ago

            ... Way of doing fixture rotation

        2. baps sniffer
          • Fantasy Football Scout Member
          • 1 Year
          2 months, 29 days ago

          These things happen, but one monster haul is a small sample size also 😉 and extremely difficult to predict. Also owning a bench player who hauls is good for tv and gives depth to squad in case of injury etc.

        3. A Fat Spanish Waiter
          • Fantasy Football Scout Member
          • 5 Years
          2 months, 29 days ago

          Your points are good ones and are well taken. Rotation is nice and all, but it doesn’t really matter when someone like Robbo is pulling consistent big hauls regardless of the opponent. Further, your analysis doesn’t speak to who will score the most points (nor is it supposed to)... but rather how their aggregate points are likely to be distributed across games based on fixture difficulty.

          That all being said, I’m going to revisit the idea of fixture rotation again because I had no idea the results were as extreme as you are showing. Plus I think the premium defenders are likely to be significantly more expensive than years past which will force us to have more 4.5-5.5m defenders in our squads.

          Lastly, I still wonder about this sort of analysis vis-a-vis the Rate My Team tool on this site. I love the tool specifically because it’s forward looking, but I don’t remember anywhere close to this level of week-to-week player variability. I guess we will see when the new season launches

          1. Markus
            • Fantasy Football Scout Member
            • 10 Years
            2 months, 28 days ago

            Remember this is only % of returns (goals/clean sheets) not total points, so if you only get 8 clean sheets a year the bulk of your points are still going to be appearance points, so fixtures not going to affect so much absolute points wise. This would be particularly the case for cheap rotation options therefore.

      2. baps sniffer
        • Fantasy Football Scout Member
        • 1 Year
        2 months, 29 days ago

        That is nice article, but it doesn't tell the whole story. For example anytime goalscorer odds should be used instead of 1X2 odds and individual assist potential calculated also.

        And sometimes we know whether to expect a lot of goals or 0-0. Usually if a draw is likely, it means probably less goals.

        1. Markus
          • Fantasy Football Scout Member
          • 10 Years
          2 months, 28 days ago

          Absolutely, it's just a shame we don't have betting odds for several game weeks. The only caveat with anytime goalscorer is that it's generally not a pure probability model, but based on betting patterns, but I agree this is a very generalised view.

          The way I see this stuff (and i don't know if it's how rmt works is): average league goals/assists/fantasy assists/clean sheets/bonus (which are most steady) > projected distribution across teams > projected fixture factoring (this + decisions about which teams go in which categories going forward) > projected distribution of those returns across individual players from that team

          1. Slouch87
            • 3 Years
            2 months, 28 days ago

            More recently most bookmakers have introduced a "build your own bet" platform. It should be possible to obtain prices for these kind of sequences now. Most of them are pricing up for a minimum of the following 2 games anyway.

  5. baps sniffer
    • Fantasy Football Scout Member
    • 1 Year
    2 months, 29 days ago

    Thanks. A lot of good work done 🙂

    One very interesting question is how to use this analysis for rotation. There are no big surprises with fixture difficulties for 3 or 6 weeks, but if we rotate not only two defenders, but also perhaps one defender and one mid/fwd some much more interesting numbers might be found.

    1. Markus
      • Fantasy Football Scout Member
      • 10 Years
      2 months, 29 days ago

      Yeah I'm sure there may be. As above, I'm wary of the other consequences of fixture rotation, ie your team is in a mess if a decent budget option emerges you want to play all the time, which is quite likely, and missing out on some returns in unlikely fixtures. But if you stuck at it over a decent period, yeah I reckon you could get some decent projected bonuses.

      To do this I've just stuck the above figures into Ben crellin's fixture spreadsheet (had to do it semi-manually to have different percentages for different tiers) based on my own projections of fixture difficulties by attack and defence for next year using the banding ranges from 18/19 which isn't too arduous a task. I'd want a bit of validation with previous seasons and smoothing across the 5 categories (eg what happens if you're 4.5* defence strength) before using it categorically for rotation projections, or preferably making public, but I think if you consider it a sense of scale that's OK. With a bit of further work the spreadsheet may be in a position to share for others to play with rotation percentages til their hearts content! I think you've got to differentiate attacking and defensive returns though so you'd want to apply it to specific player projections rather than just total points (hence rmt a great tool)

      1. baps sniffer
        • Fantasy Football Scout Member
        • 1 Year
        2 months, 29 days ago

        RMT is a nice tool, but I am not always happy with it's projected points, at least in short term. So perhaps creating own model and using both team stats and own watchlist projections might be the way to do it. RMT does also have some limitations when planning future moves.

        There is a lot of work to do, but following only teams with decent fixtures and not too many players might help.

        I think it is always better to plan and have also plan C (since plan A and plan B fail too often) than not have alternative strategies preplanned at all.

        1. A Fat Spanish Waiter
          • Fantasy Football Scout Member
          • 5 Years
          2 months, 29 days ago

          I’d actually love to look behind the curtain to see what’s really going on there. The thing is though, it’s the best forward looking tool I’ve seen, and while I have never seen an audit of the tool’s results, I trust it far more than my ability to generate the numbers on my own.

          But you are right about one thing... it’s slow to react. Look no further than the fact it rated Pogba highly at the end of the year, despite terrible team and individual form. I suppose that just means that the eye test is important too.

          1. baps sniffer
            • Fantasy Football Scout Member
            • 1 Year
            2 months, 28 days ago

            Pogba for example was not only about eye test. His position changed to worse and underlying stats dropped, I believe David Munday wrote articles about that.

            The job is difficult though. It is not so easy to predict when someone is in form or not. Trying to "nail it" would also mean serious failures.

            Yet I believe that if numbers are "correct", at least according to your own opinion you can use them in decision making process. The problem is the number of unknowns. What will the coach think, who will start, how long will someone be in form etc.

            One of the problems with the tool is that projects probability to play. For me it would be better without out that. Guessing how Pep roulette is programmed is annoying. I would also prefer to judge myself who is fit to play and who is not.

  6. Triggerlips
    • Fantasy Football Scout Member
    • 9 Years
    2 months, 29 days ago

    Great article, only potential issue is how stable are attacking and defensive returns from one season to the next.
    In that respect it is similar to all analysis of this type, great at showing what has happened but not a guaranteed predictor of what will happen.

    1. A Fat Spanish Waiter
      • Fantasy Football Scout Member
      • 5 Years
      2 months, 29 days ago

      No question. As we say in the finance industry, past performance doesn’t indicate future results. However, this analysis seems to show that any one player’s week to week performance is likely to (on average) vary considerable with the ease or difficulty of the fixtures.

      I think most of us intuitively understand this to be the case...I was just surprised by how large the variation, particularly for anyone who doesn’t play for City or Pool.

    2. FPL Virgin
      • Fantasy Football Scout Member
      • 3 Years
      2 months, 29 days ago

      The man, the legend! Great to see you on here.

      Do you still plan to release your FPL book for free? Would love to give it a read in the offseason.

      1. Christina.
        • Fantasy Football Scout Member
        • 9 Years
        2 months, 28 days ago

        Free?

        You need to demand to get paid to read it.

    3. Markus
      • Fantasy Football Scout Member
      • 10 Years
      2 months, 28 days ago

      For sure. But I'd like to compare to other seasons as my gut is that (obviously for different team who are strong in different seasons) it would paint a similar picture as the affect of different fixture difficulties and smooth out some of the curves. So in that sense, I'm looking at a model that can be applied as general rule (like 88% goals result in an assist) which won't bear out exactly in the forthcoming season, but is a best informed guess. What it won't do is categorise teams or correct fixture difficulties which will of course change from season to season - that is the human/wider statistically informed part of the projection, and obviously the most important one in scoring real fantasy points too!

      1. baps sniffer
        • Fantasy Football Scout Member
        • 1 Year
        2 months, 28 days ago

        There is one problem though. Stats like this do predict the past if categories are decided by previous results. And a team like Wolves doing well against better teams is a problem also and should probably be handled in a different way (discard?).

        Interpolating these results is good enough for me and I have my doubts that more work would change results too much. Yes, category 5 teams had a great season, but I believe this will predict the future more than findings from previous seasons.

        1. Markus
          • Fantasy Football Scout Member
          • 10 Years
          2 months, 28 days ago

          How much do you think teams that overperform vs top sides (and by definition therefore underperform vs worse sides) is translatable from season by season? I see this more as part of the unpredictable element of football rather than anything else, but that may be too simplistic.

          1. baps sniffer
            • Fantasy Football Scout Member
            • 1 Year
            2 months, 28 days ago

            Tbh I don't. Yet for Wolves it was predictable pattern at in the spring when the trend did continue. I think the reasons have been found out.

            So perhaps the reason is that counter attacking teams in mid table may be more "fixture proof" than others. They do well against better sides, but may be in trouble against lesser opponents.

            I haven't checked out the stats, but checking out individual attacking assets over more seasons (Jamie Vardy is interesting now) is worth extra study. I believe he (and Jimenez) has less attacking ppoints correlation vs opponent. On the other hand Callum Wilson wasn't good vs top sides.

            My guess is that leads to strategy that you could play Vardy and Jimenez against any opponent (besides top2), but Wilson is the
            player you need to have for good fixtures only. (And I am of course not suggesting to have these exact players or that 3 fwd:s is a good idea). Yet it would be convenient to have enough core players that are more fixture proof and then be able to change others.

            I am happy to share some of the work 🙂

            1. baps sniffer
              • Fantasy Football Scout Member
              • 1 Year
              2 months, 28 days ago

              Sorry for those typos, writing in the sun with mobile phone 🙂

            2. baps sniffer
              • Fantasy Football Scout Member
              • 1 Year
              2 months, 28 days ago

              And this is the funny part in fpl that I love. Nerds or geeks know their math, but don't understand enough of the game. Eye test usually doesn't help them. Also converting meaningful data to good decisions is too difficult for them.

              Football enthusiasts perhaps understand the game, but they usually have too short memory and the math part is too difficult for them.

              So in theory those who master both sides of the game should have quite big advantage. Yet unfortunately luck does play very big part also. There will always be coin toss situations and also unexpected events do happen.

              1. Andy_Social
                • Fantasy Football Scout Member
                • 7 Years
                2 months, 28 days ago

                There are always non-mathematical further issues to factor in. For example, you mention Wolves. For me, their participation in the Europa League each Thursday evening in the first few weeks of the season is a game-changer. Look how it affected Burnley last season. So as much as Jiminez, Doherty and latterly Jota served me extremely well last season, this time I'm starting Wolves-free.

                1. baps sniffer
                  • Fantasy Football Scout Member
                  • 1 Year
                  2 months, 28 days ago

                  Yes, and this makes it very difficult to analyse old data. It is impossible to remember all circumstances afterwards.

                  One unforeseen situation is VAR in PL. Has anyone seen study how it would have affected last season? My gut feeling says that we should see more penalties especially against big teams.

                  (But thanks for reminding me not to get Jota in 😉 )

  7. FPL Virgin
    • Fantasy Football Scout Member
    • 3 Years
    2 months, 29 days ago

    Superb article - thank you. I like to think I'm pretty good with numbers and stats but I could never write an article like this. Ultra high level stuff.

  8. 11 smelly shirts
    • Fantasy Football Scout Member
    • 1 Year
    2 months, 28 days ago

    "A rubbish team doesn't suddenly get great - and vice versa..." May I assume that excludes Man United? :):):)

  9. A Fat Spanish Waiter
    • Fantasy Football Scout Member
    • 5 Years
    2 months, 27 days ago

    All of this got me thinking about rotation, which I realize is topic on which many mangers have strong views (mostly against). Most of the discussion focuses on home/away rotation for positions like GK or DEF... but what about fixture rotation?

    It turns out there are two sets mid-table teams with outstanding fixture rotations this season. Those teams are Everton/Palace and Wolves/West Ham. In fact with either rotating pair, you would never face a top six side all season.

    At the same time, there are a few pairs, most notably Villa/Norwich that would make horrendous rotating pairs.

    This has definitely got me thinking about doing two premium defenders, two rotating lesser defenders and a 4.0 defender, while going 3 at the back as my default.

    For those that care, my method was to assign a value to each opponent (top 4 teams are worth -2 points, the next set -1, and so until you get to the bottom teams which are +2). Then you simply have a spreadsheet chose the best team of the pair and add up the score.

    As an additional aside a “perfect” score using this method (which I fully stipulate is somewhat arbitrary) is 44. The EVE/CRY pair scores 43. WOL/WHU is 41. NOR/AVL is 19.

    1. LJ
      • Fantasy Football Scout Member
      • 4 Years
      2 months, 27 days ago

      I was looking at EVE/CP very early on, not just for a GK rotation but for a DEF home/away rotation. The fixture list for them looks very very promising.

  10. stat
    • Fantasy Football Scout Member
    • 3 Years
    2 months, 25 days ago

    Great article mate!

    I wonder how this translates to individual players (mainly mids + attackers) as there are some players who tend to do better in tougher games (Jimenez) than more favorable fixtures.

    I can't really think of any example off the top of my head, but I'm sure there are players who do better against easier opponents and players who do better against tougher opponents even though both players play for the same team.

    But I'm wondering if that's just down to variance, or whether it highlights that perhaps a player is better suited to playing on the counter.

    1. Markus
      • Fantasy Football Scout Member
      • 10 Years
      2 months, 25 days ago

      I think variance is a big part, see this article https://mathematicallysafe.wordpress.com/2018/07/08/fpl-analysis-the-impact-of-fixtures-on-player-performance/
      The fast track bully feature would be an interesting one to test on counter attacking strikers (vardy and Jimenez the most obvious two) although wolves debatable based on this https://www.premierleague.com/stats/top/clubs/goal_fastbreak?se=210

  11. Yellowbee
    • Fantasy Football Scout Member
    2 months, 6 hours ago

    Great article, thank you sp much!