Football analytics: when football meets science

Written on 12 November 2017, 09:51pm

Tagged with: , , ,

I wrote a piece about football analytics in Romanian: when football meets science. It was one of the articles I really enjoyed writing and it took me over 10 evenings to do it.

Here are the top level details:

Football analytics is all about using data about previous events in order to have an indication about the outcome of future events.
It is not new: it started somewhere in the ’50 and one of the first coaches to use it was a Russian trainer called Valeri Lobanovsky, in an era where a computer was taking up rooms.
I found a correlation about the DIKW pyramid and the usage of football data:
– Data – numbers and metadata collected using manual operators, tracking devices or video tools
– Information – when data is put into context. One indicator that recently became mainstream is the ‘expected goal‘ (xG) – a percentage associated with every shot based on previously aggregated data
– Knowledge – when information is combined with previous experience. Example – aggregating information about indicators like xG (xG for, xG against, non-shot xG, xG difference)
– Wisdom – using previous levels to take strategic decision enabling competitive advantage.

The first two levels are for the football fans, media writers and TV pundits.
The last two levels are for the professional football clubs and for the betting companies. This is where the football analytics takes places and these levels can give indication about future events.

A few examples of football analytics:
1. transfers: before any transfer, the targeted player is analysed from a few perspectives: tactical, physical, technical. The modern clubs are using players databases with custom criteria in order to maximize their match rate.
2. injury prevention: by tracking the way a player runs and measuring how long his feet stays on the ground, one can evaluate the player tiredness
3. predicting outcome of future events by calculating and maintaining a club index (ex. fivethirtyeight.com)
4. penalty shoot-out: statistics showed that the team shooting first has a 20% advantage over the second team. The football governing bodies realized this un-fair advantage and recently changed the order of the shoot-out (now ABBA instead of ABAB)

In the end, football remains a random sport. Using analytics can give indications, and make the clubs better understand some questions, but it cannot (yet) give definite answers. As long as football is played by humans, the human factor will play its part and will keep football random and enjoyable.


The graphics on Fifa 16 are something else

Leave a response