Thursday, 30 July 2015

The road to FPL 2015-16 (part 1)

This post is the first in a 3 part series leading up to kick-off in the EPL and, of course, the FPL season.

While on vacation (and finally with some time on my hands), I had the opportunity to dabble with the FPL data I pulled last season, and found a few interesting things that could help forming the initial squad for this campaign. The series will get increasingly math-heavy as it progresses, but I'll try to keep it as light as possible.

  • Football is like the weather
    • This season is going to be more or less like the last one
    • Data from last season can be used to form this season's teams
  • Averages are fine (for now)
  • Relegated teams are interchangeable with promoted ones
    • In place of Hull, read Bournemouth
    • In place of Burnley, read Watford
    • In place of QPR, read Norwich
  • You like looking at line charts
Picking a fantasy football team requires you to hit several sweet spots (as you'll see), and still pray things on the pitch turn out the way you want. But hey, let's leave as little to chance as possible, eh?

The plum ties

Pick apart the fixtures that would yield points for your team. Doing this is essentially ‘frisking the fixtures’, as Fantasy Football Scout calls it. Teams are ordered by the average points they allow per game per player (If you think the order looks familiar, take a peek at last season’s league table)

Fantasy points allowed per game per player

Tottenham aside, teams did better at home and worse away, as conventional wisdom suggests. The scale of the difference in performance varies by team, however. The likes of Crystal Palace and Sunderland showed consistent leniency regardless of venue, while the extent of Newcastle and Southampton’s generosity depended on where they played. While Everton were an easier opponent on average than Arsenal, it's noteworthy that Everton away was an easier fixture than Arsenal home. Enough variation to look at venue as well as opposition rather than just the black line.

The bankers

You know the pick of the plum ties, but you'd still need to be sure which pool of teams to select players from. Teams in the below chart are ordered by the average points they score per game per player.

Fantasy points scored per game per player

You would notice a minor change from the previous plot - Spurs conformed to the home-away belief when it comes to points scored, while Palace did not. This plot also confirms the idea that a mere overall number for each team will not suffice. Chelsea posed more of a threat away from home than Everton did at Goodison Park, and the teams with a noticeable difference in performances here remain more or less the same as compared to points allowed.

The roles

Now, how about positions on the pitch?

Fantasy Premier League points by position

The best performers are the keepers! This shouldn't really be surprising - Goalkeeper is one fixed role, while there are different types of defenders, midfielders and forwards, some of whom are more likely to score points than others. This variation, which we'll explore in more detail in the next post, is likely pulling down the numbers for the outfield players, and also puts paid to the use of averages for these positions.

So, we’ve got a broad idea of:
  1. Which fixtures are actually good
  2. Which teams are actually good
  3. Which positions are actually good (kind of)
This leaves us with more questions:
  1. Who are the good fixtures good for?
  2. Who is good to pick from the good teams?
  3. Who are the good players in these positions?
To answer these questions, we'll need a marriage between teams and positions to give us the clarity we need to answer these questions. Let's break down Manchester United's generosity last season by position, for example.

Points allowed by Manchester United

One thing that should jump out at you is the fact that travelling goalkeepers did almost twice as well as the defenders at Old Trafford. This difference is far less pronounced when the Red Devils travel, meaning Manchester United’s generosity towards defenders varied wildly based on whether they played home or away.

The story of variation persists in another example, when we break down Southampton's performance by position. Their strikers provided wildly varying returns based on venue – they were the best Saints assets to possess at St Mary’s and by far the worst away from home.

Points scored by Southampton

Clearly, in order to pick a team, we need to look at a combination of:
  1. The team under consideration
  2. The opposition
  3. The venue
  4. The positions
    • The players who aren't goalkeepers
That sets the agenda for part 2, we will be gathering team selection wisdom from the marriage and look at some FPL trivia.

Anyone who wishes to access 2014/15 FPL data can do so here.