In Search of The Best Captaincy Indicators

Following on from my analysis of this season’s most explosive and consistent Fantasy Premier League captaincy options I’ve decided to take a closer look at the various indicators we can use to assess a good armband pick. This includes a closer look at price, form, fixtures, peer consensus and the various other statistical evidence at our disposal.

THE METHOD

I weighed each of the following indicators equally, giving them relative strength of captaincy scores from 0-4.

Instead of ranking them based on the competition among other players from week to week, I kept the numerical values constant across all Gameweeks for two reasons: I wanted to create baselines of captainable performance expectations in case an FPL manager values consistency, and I wanted to show relative outliers in any given category if an FPL manager craves explosiveness more.

The first two, of price changes and the captaincy poll, look at what would happen if we simply followed the masses, either by their transfers or their votes:

Price changes 

When pulling the data off the FPL player pages into a spreadsheet, I noticed the Teams Selected By and Price columns in my graphs.  I wondered how accurate the purchasing habits of the FPL masses might be in predicting a haul, so I weighed Net Transfers In (NTI) and price changes with the following values:

1 – at least a 10% increase in total owners but no price change

2 – increase of 0.1

3 – increase of 0.2

4 – increase of 0.3

Sometimes prices change from Gameweeek to Gameweek based on Friday night/Saturday morning transfers, so some of these are skewed and do not reflect a massive surge in ownership.  This did not happen too often with the players we look at here, however.

The Captain Poll

1: 10-19% or top five

2: 20-39%

3: 40-60%

4: 60+%

The margin of error here is certainly significant, but I don’t think we should discount that it does have the ability to capture a general sense of Fantasy Football Scout community’s mood regarding the armband.

The next two are the classic, bedrock indicators for many FPL managers –  form and fixtures

FPL Form

1: 5-6 PPG

2: 7-8 PPG

3: 9-10 PPG

4: 11+ PPG

Fixtures

FFS Season Ticker color ratings, 1-4 based on shades of blue

The last two are more statistically based, with my homemade one as well as FFScout Member’s area Season Ticker.

Statistics 

A bit of a leap here, but I weighed three of the underlying stats — Shots on Goal, Chances Created, and Big Chances – in the following manner:

SoG (x4 Forwards or x5 Midfielders)

CC (x3 both)

BC (x2 Forwards or x2.5 Midfielders)

We often focus on goal threat with underlying statistics, and rightly so, but I also wanted to give credit to the assist makers who are creating chances.  The raw data is taken from the Captain Sensible articles and just a basic Goal Threat table I’ve made as well.

As you might be able to tell, the Shots and Chances are valued as if they are converted FPL goals and assists for the two attacking positions (Big Chances are weighed as an extra half-goal for each); in essence I wondered how many points a player would earn if he were converting every shot and his teammates were scoring from every one of his passes.

I don’t pretend that this is the most accurate metric and I fully acknowledge I am wading in the shallow end of the soccer statistics pool that the xG folks understand much more deeply, but it is nevertheless a baseline for the underlying stats.  Assuming a 20% conversion rate, a score of 48 over the last four matches would mean 3 raw FPL attacking points/match, which when coupled with the resultant BPS would be about a 6 point gameweek.

1: 50s

2: 60s

3: 70s

4: 80s

Rate My Team 

1: 4.75-5.50

2: 5.50-6.00

3: 6.00-6.50

4: 6.50+

Total

The sum of all the categories, given three shades of green based on how strong the indicators were or how many different categories indicated support for the captain’s armband.

FORWARDS

Diego Costa

Mr. Consistency doesn’t seem to have too many patterns to follow for his two Explosive performances, nor do his most likely captaincy Gameweek result in any discernible uptick in production.

Ignore the hype, give him the armband, count your 10-20 points, it seems.

Romelu Lukaku

One area to concentrate on with him might be Fixtures, as his strong start has been matched by a pretty favorable set of fixtures for Everton.  His Dynamite haul and one of his Explosions have come when indicators of a good fixture are there, and he delivered a 12 pointer during his strongest fixture, the match against Middlesborough when indicators across the board highlighted his potential.

The RMT tool also seems to gauge Lukaku well: 54 points from the five times he’s been a top pick, versus 15 points from the five times he hasn’t.

Sergio Aguero

Extremely inconsistent has been the story of Aguero’s season, with little rhyme or reason since Gameweek 3.  One area to consider is how the masses favor him: when he is the darling of the FFScout Captain’s Poll as a runaway leader, he’s delivered all but once, including his two Explosive hauls.

Of his four most widely indicated Gameweeks for a strong performance, he has delivered three times, but his only Dynamite performance so far occurred when he was at his least popular captaincy point in the season.

However, this was due to a double benching over the previous week and speculation he was out of favour with Pep Guardiola. With that notion dispelled a fixture such as his away clash at West Brom, where he earned 16 points, would more likely see him nearer the summit of the Captains Poll.

MIDFIELDERS

Eden Hazard

A double digit haul on the week his indicators most strongly suggested it, might reveal a consistent performer, but his subsequent blank on Gameweek 4 is just one example of what appears to be a very inconsistent captaincy candidate.  Only home matches seem to provide any semblance of a trend or pattern in his performances.

It remains to be seen whether the Chelsea’s Gameweek 7 switch to a 3-4-3 formation will continue to make him a strong candidate. He has delivered points returns in all bar one of the league games where this set up has been deployed.

Kevin De Bruyne

The stars aligned for him in Gameweeks 5-6, and he delivered quite nicely.  Maybe one of the surer picks in this whole list.

Phillipe Coutinho

Roberto Firmino

Sadio Mane

Checking in on the Liverpool trio shows the varying levels of popularity they possess.  Coutinho seemed to be more consistently meeting expectations before this most recent international break.  Firmino has been a bit more frustrating for owners trying to catch his hauls, with the Gameweek 6 blank against Hull one of the more shocking returns of the season, while his Gameweek 11 master class against Watford rewarded the faithful.

Mane seems to be a Home fixture star, but what is perhaps most interesting about him is how little attention he received across the spectrum of indicators.

Alexis Sanchez

Theo Walcott

Wrapping up with the Arsenal duo shows just how maddening a captaincy failure can be.  Sanchez had three successive weeks where all signs seemed to point to at least an explosion if not a Dynamite performance, but his captainers were treated to a sequence of 3, 5, 3.

Likewise, Walcott boasts the most emphatic week so far, a Gameweek 9 clash at home against ‘Boro where every indicator was at least strongly on his side.  His clean sheet point in a 0-0 sickener provided little solace for owners.

TRENDS

While the above numbers try to explore how the players are performing, are there any indicators that might be seen to be more or less effective so far this season?

For a number of different reasons – the subjectivity of the weighting/ratings, small sample size, and the overall speculative nature of this exercise, to name just a few – trying to answer this question is difficult and potentially quixotic. But here goes.

I looked at the total number of points earned in each category and divided them in two ways:

Points/weighting — amount of ratings points, adding up all the 1-4 weights (to take a stab at an idea of Explosiveness, very shaky methodology here)

Form 4.40

RMT 4.37

Fixtures 3.58

Poll 3.36

Price Changes 3.20

Stats 2.80

Points/indicator — number of instances when any indicator highlighted a player, regardless of whether it was a 1-4 (to take a stab at an idea of Consistency since the indicators were trying to tell us something, a bit firmer ground here since it isn’t rating or evaluating the quality or weight of the recommendation)

RMT 7.78

Fixtures 7.67

£ 6.47

Form 6.41

Poll 5.93

Stats 5.54

Ironically, RMT and the Stats are at the top and the foot of the table, but that likely has much to do with the statistical prowess over the long haul of the Members’ Area’s defining metric when compared to a cruder, more selective look at a few underlying stats.