Second Half of the Season Trends Analysis

This is the first in a season-long series of articles from FPL veteran, poker aficionado, former professional journalist and all round good guy Nick Wright (follow him on twitter here).  

Idly browsing the data set from the 2019-20 season, I realised we might have missed something.

While we always need to be wary when slicing and dicing an already small sample size into an even smaller one, I decided to look at FPL data exclusively from the second half of the season to see if there were any trends that we might have missed, versus looking at the season-long data set from 2019-20.

Here are some things I found:

Manchester United – Trending Nowhere

The newspaper narrative of Manchester United after project restart was that they blew teams away with their array of attacking talent. In FPL an attacking double up was standard and the triple-up was common.

A lot of this has been put down to “the Bruno Fernandes effect” – but the numbers paint a different picture.

The numbers say that Bruno Fernandes had a negligible effect; United’s xG differential barely changed after they signed him;

 

Without Fernandes

25 games

xG – 37.9 (36 actual goals)

xGA – 22.2 (29 actual goals against)

1.51 vs .89 (differential of .62 per 90)

 

With Fernandes

13 games

xG 21.6 (30 actual goals)

xGA 13.7 (7 actual goals against)

1.66 vs .1.05 (differential of .61 per 90)

(All figures from https://fbref.com/en/squads/19538871/20 … ier-League)

 

The ‘Fernandes effect’ was that United’s attack had a slight uptick, but at the expense of their defence.

The numbers suggest that United ran red hot in front of goal and we need to be wary that we don’t fall for recency bias here when selecting our initial squads. Their ‘with Fernandes’ goal total works out at 88 for an entire season – only Man City have exceeded that figure in the past five seasons.

So, although Bruno was seen as the catalyst that helped kickstart Manchester United’s season, their xG differential barely budged. Stats sceptics will tell you his importance has been overblown whereas those who trust the eye test will tell you that his arrival turned Manchester United from a team who could only attack on the counter into a team who now had multiple ways to attack games.

Nonetheless, whichever way you slice it, 13 games (where Bruno had a lot of penalties) is not enough to know for sure.

If we extrapolate their 1.66 xG per game to an entire season it pegs them as the fourth best attack, miles behind City and Liverpool, and still behind Chelsea.

They did have the third best defence for xGA last season, perhaps we’re looking for value at the wrong end of the pitch?

 

Tottenham – Trending Downwards

Spurs fans appear to be pinning their hopes of Mourinho doing second season Mourinho things. He’ll no doubt make an example of a high profile player and be fun to watch in press conferences.

But – whisper it quietly – Spurs haven’t been good for a season and a half now.

The poor performances at the back end of the 18/19 season were masked by a Champions League final appearance.

In the second half of 18/19 under Pochettino, Spurs ranked 11th best in attack and 8th best in defence for xG and xGA – more or less league average.

The slide continued into the first half of last season, which led to Poch’s dismissal. Surely miserly Mourinho at least fixed the defence though?

No. It’s gotten worse.

29.55 xGA in the second half of the season, 12th best, and 24.58 xGA in the first half of the season, 9th best.

“The truth is, Spurs metrics project their defence as mid-table, but we’re being asked to pay 5m+ for defensive assets who have the same prospects as a host of cheaper assets from teams like Southampton, Brighton and Burnley. Whilst for the same price we have routes into the superior Wolves and Manchester United’s backlines.

 

West Ham Attack – Trending Upwards 

Hats off to David Moyes. He turned West Ham from relegation certainties into mid-table fodder.

Using Understat’s xPts model, West Ham were 19th over the first half of the season. Moyes took over on 30th December, and during the second half of the season that same model ranked them 11th.

Whilst they improved at both ends of the pitch, it was their attack that really prospered.

xG vs xGA First Half of the Season

20.77 vs 35.36

1.09 vs 1.86 (-.77)

 

xG vs xGA Second Half of the Season

27.52 vs 30.34

1.44 vs 1.59 (-.15)

A fluid front three including Micahli Antonio and Jarred Bowen saw West Ham’s attacking numbers go through the roof, with 155 SITB compared to 132 over the first half of the season.

Moyes also realised that Mark Noble +1 doesn’t constitute a Premier League quality midfield, and switched to a three man central midfield for much of project restart.

 

Sheffield United Defence – Trending Downwards 

If you were worried about investing in a Sheff Utd defence without Dean Henderson, the truth is you should’ve been worried even if they had retained his services. The stats make grim reading for Blades fans worried about a second season slump.

Over the first half of the season they ranked fifth, with an xGA of 22.08, this despite leading the league with six defensive errors that led directly to a goal. The second half of the season, not so good. Their xGA of 29.96 ranked them 14th best and their Shots In the Box conceded increased to 149 from 117. Brighton, Palace, Southampton and Everton all put up better defensive numbers.

Putting this defensive malaise down to the conditions they faced due to project restart is fanciful at best. Villa, B’Mouth, Palace, Burnley and Southampton, who all operate with small squads, all performed better.

 

These trends have certainly influenced my own squad selection for the new FPL season and I’ll be keeping a close eye on the four teams above to see if they continue.