Thursday, 5 February 2015


Week 29

Shots For & Against P/90

A motion chart showing shots for & against per 90 minutes. 

Press play for motion. Green=top4. Red=relegation places. Blue=league average.

Liverpool (improving) and Arsenal (regressing) both look in better overall shape for the Top 4 in terms of shots for and against than Manchester United do.

If team x is above and to the left of the blue dot then said team is below league average in both shots for & against. If team y is below and to the right of the blue dot then said team is better than league average in both shots for and against.

Good/Lucky Chart: SoTR With PDO

Paul Lambert has finally been dismissed from his position as manager of Aston Villa. Why? His Villa team have never cracked the 40% barrier in terms of shots on target. Since 2009/10 thirteen teams have posted season-long sub 40% SoTR numbers, seven of those teams were relegated. 

Don't let the fact that only thirteen teams have ever posted a SoTR (shots on target ratio) below 40% escape you. Villa are a historically bad team right now, at week 25, in terms of their shots on target numbers. 

A quick word on Villa's PDO: Right now it's in the gutter at 87 but there was once a time (after game week 4) when that now terrible PDO number was at the lofty heights of 119. Why is this significant? Randy Lerner, Villa's owner, decided to award Paul Lambert with a new 4 year contract after week 4 when the club sat comfortably on 10 points. Villa had 10 points from 4 games having taken 38% of the shots on target and scored 80% of the goals. Villa's PDO (& some score effects) were the difference. 

Villa's Goal% number was never going to sustain because the PDO was always going to regress back toward the mean. Unfortunately for Villa and Lambert, the PDO has not only regressed toward the mean, it has shot past it, through the fucking floor and into the basement (PDO is now 87). Yet Villa's SoTR is exactly the same as it was at week 4 as it is at week 25, 38%. Lambert has been fired for making Aston Villa simultaneously bad and 'unlucky'.

Not only can PDO kill coaches but it can also trick owners into stupidly offering new contracts for personnel who don't deserve such offers. 

People may not like PDO but one of its main strengths is to identify outliers and the unsustainable and identify Villa as an unsustainable outlier it did. 

People also don't like analytics much but you know what analytics can do? They don't give magic answers or bullet-proof solutions; analytics can prevent the stupid decisions and obvious mistakes. And in this case it would have prevented Villa's owners/decision makers from making both stupid and oh-my-fucking-god expensive mistakes.

Paul Lambert was sacked just 21 Premier League games after signing his new 4 year deal. Head-shaking stupidity from Lerner to award that deal after a good start to the season which was powered by some unsustainable conversion percentages. All Lerner had to do was wait a little, see how Villa looked at the halfway point, see if the team had really changed and improved. You should of waited, Randy. Now look what a mess you're in.

Scoring% & Save% Regression

Blue = League Average | Size of the bubble is Points

Aston Villa are worth clicking on and tracking as the season progresses. We can see the size of the bubble shrink, we can see the save% regress and we can see the scoring% move sharply from right (good) to left (bad).

The amazing thing about this chart is how the spread of teams and the rates at which they score/save shots on target tightens up so quickly. This is the regression of the percentages in motion.

Shots p/90 & Key Passes p/90

Ozil & Lamela are posting very similar 'basic' numbers here, and that's pretty weird to me. Not sure how much we will learn about Ozil and the end of what will be another injury-riddled season, but it feels like there is a not insignificant gap between reputation and actual countable performance.


Goals p/90 & Assists p/90

We have seen the process for these selected forwards, now here are the results in goals & assists p/90 form:

Shooting% & Assists Shooting%

And this is how those guys turned their shots and passes into goals and assists: Shooting% & Assist shooting % (assists/key passes).

As you can see the spread of shooting % and assists shooting% is all over the place. These numbers have, and will continue to, regress as the season grinds on. 

Jonathan Walters, though.

No comments:

Post a Comment

Note: only a member of this blog may post a comment.