Page Nav




Breaking News


Stats infesting football twitter: The Moneyball effect

T his is a challenging article for me to write because I don’t believe stats should be used on football twitter. I believe we should all ...

This is a challenging article for me to write because I don’t believe stats should be used on football twitter. I believe we should all trust our observations and feelings and be content with that, without this obsessive need to be a performance analyst. We fell in love with football because of feelings and emotions, not points-per game metrics.  

Statistics are quantifying the beautiful game to a sterile wasteland where feelings are replaced by metrics and observations are worthless without a graph

But rationally it makes sense to use statistics. The reason why it makes perfect sense to use statistics is that observations are extremely unreliable. Elite coaches only accurately recall 40-50% of details from a goal they’ve observed and only accurately recall 20% of details from general play. Similar studies have been conducted regarding topics such as eye witness statements and people rarely accurately recall 50% of the details of a crime they’ve witnessed. There using statistics is perfectly rational and better than trusting yourself, according to the stats. 
We can’t draw many conclusions from statistics. Statistics can only give us an idea of how important something is but are very limited in terms of validity. Statistics are only significant when the statistic has a large correlation with the match result.  For example, the number of shots is more significant the number of passes made but very few statistics correlate with the result of the match. However, there’s a small correlation between possession percentages of possession-based teams and the result. But even this is flawed as teams with the better players will have more possession the majority of the time and therefore they’ll win more games due to having better players. 
Stats should be used to justify an opinion, not form one. For example; if watching a game you think a team’s high press has been successful you can look at stats for tackles, duels and interceptions of the attackers and pass completing rates of the defenders and goalkeeper. And even if these don’t correlate that’s fine. You think they have been successful and that’s enough. It may be that the press was successful because the defenders had to play long balls forward and the defenders were able to dispossess the attackers as a result. 
To provide more validity values such as Xg were invented. This calculates the predictability of a goal to be scored in certain areas in certain situations. This should, in theory, calculate how many goals a team should score. 
Statistics are completely objective, the most common stat in football is possession rate and more recently expected goals (Xg) and even expected assists (Xa).  When Guardiola’s Barcelona were kings of Europe everyone went mad for possession stats, pass completion and thus we entered the age of possession analytics. Now we’re progressing into the wonderful world of expected goals (Xg). This is now the poorly interpreted and awfully applied statistic. Xg is an average. Therefore it requires some to be above it and others to be below it. Low and behold Liverpool outperform their Xg because they have one of the best front 3s in world football so of course they’re clinical.  

Facts are solid but statistics are pliable 

Stats should be left to the performance analysts. Professional clubs will record and measure everything possible. If you can think of it someone else has thought of it and someone else is measuring it. And this probably applies to the academy as well if the club has the resources. The analysis process will probably look like this:
  1. Record the match 
  2. Download the footage onto a laptop 
  3. Use software to code the match according to the club specification
  4. Highlight any clips of the match where the team did something that can be measured. Ranging from switches of play, final third entries, crosses, recovering into shape and anything else they can think of. 
  5. Highlight individual clips of players to send them their game highlights with a few points to improve on. 
  6. Send all the clips to the head coach who will then feedback what they feel necessary to the players. 
  7. All clips get saved and the process will start again for the next match. 
This example is excluding the physical data that would be recorded by GPS vests. 

Descriptive statistics exist to simplify with a loss of nuance and detail 

The main reason why I dislike the use of statistics in football is that they can’t quantify decision making. The difference between the very best players in the world is decision making. A player can have 110 passes per game but if in each of those games they miss the chance to play someone through on goal this doesn’t get recorded. You can’t quantify the correct decisions made by a player because the notion of ‘correct’ is subjective and stats cannot be subjective. 
The growth in stats on football twitter is due to Liverpool’s success and there well known successful use of stats in recruitment. The Moneyball effect in full swing. However, Moneyball was used in baseball which is a similar game to quantify due to the reduced decision making. There are a pitcher, a batter and some batters on bases but they’re the only moving parts. The pitch throws and if the batter hits it they can only go to first base, then second and so on. The man decisions to be made or when to stop at a base and what base the fielders throw the ball to. 
Another reason why I don’t like stats on football twitter is they cannot measure what occurs off the ball. On average a player won’t spend more than 2 minutes with the ball in an entire game. Yet stats only measure what a player does with the ball. It is possible to measure off the ball actions but these are far more limited in terms of usefulness, restricted to distance covered, sprints, heat maps etc. You can’t quantify when someone is in a good position off the ball. It is impossible to count the number of times when a player is in the right position as this is subjective and statistics cannot be subjective because this defeats the point of statistics. Off the ball, movement is closely linked to decision making so these two are coupled.  

To conclude I don’t believe stats should be used on football twitter as they fail to measure decision making both on and off the ball which are the most crucial aspects to football. A further frustration is due to the sheer number of stats available even a drunk could use them for support to stand, creating an illusion of stability but remove them and they will fall down. 

Written by - @benbell98

No comments