Thread by @babuyagu

Data is invaluable. Seriously. It is unbiased and massive. One person simply wouldn't have the time in their life to watch every game of football every weekend to build up a comparable 'data set' in their own head to draw fair comparison of players.

The problem is always in how

https://twitter.com/estreetbruno/status/1427290089445629955

good your data is and how good the person is at using it.

You can give the best, most accurate, most comprehensive set of data in the world to a fucking idiot. If they are setting out with the sole purpose of proving Sol Bamba > Virg they will find a way to dig out a few things

to prove themselves right and ignore everything that proves them wrong. It is just an exercise in confirmation bias, with cherry picked data.

Essentially you are only seeing that person's opinion. If they told you it without data, it wouldn't be any better than with data.

However, at least with data you are seeing what they are basing their judgement on. And then if you have access to the data you can pick it to pieces.

Without data - it is just an opinion. Which ultimately ends up in two people with entrenched opinions shouting "no, I AM right"

One of the things you learn when working with data - and I mean you get paid based on your ability to use data... is that you will get fired, and possibly sued... if you don't find a way to remove your opinion and bias from your work.

A top club doesn't give a fuck what a data scientists opinion on Doku is. If they like watching him or not. It has no value to them. They have scouts for opinions - and will trust their trained eye overall.

What they want is the cold, unopinionated data. They want to know before

they splunk £50m on a guy what all the data of hundreds of games says. To fill in all the gaps between those 15 scout reports and clipped videos don't say.

And this is also something people miss with FSG and them 'not really knowing football'. It doesn't matter. In fact it can be counter-productive if they do - or worse, think they do

If they can get the best people in throughout the organisation at their jobs and then hire someone

who excels at receiving information from all those people, interpretting it, deciding what is valuable vs what isn't, then making the right decisions on it - and they empower them to do so - then you have a successful formula right there.

Whereas if you get someone who thinks he knows football - like a Jack Walker - who is telling you that Zidane is no better than Sherwood and you don't need him. Or selling Makelele because they don't know what he does. Or a Christian Purslow who is playing Football Manager in

real life and just deciding what is best rather than Rafa Benitez - you are fucked. Because you have the person with the most knowledge and experience at the job being dictated to by a casual.

Here is an example of this in action. The NY Times did a piece on Dr Graham - who heads up this department at Liverpool. He sat watching the game against Leicester and was cheering Keita on. When asked his opinion on his performance, he didn't give one.

https://twitter.com/babuyagu/status/1427296509142487050

Instead he said he would need to look at the data. Dr Graham isn't paid for opinions, he is paid for data-based analysis. He trusts his algorithms and Liverpool's data. Not his own opinions which are loaded with bias.

And this is the thing maybe people don't understand.

If you told them a club signed a player based on the recommendation of a data scientist who never saw them play live in the the flesh they would think that is insane.

But that is the process. They can identify the player. Clip together videos and send it up the chain. Then a

host of different scouts will watch dozens of games of the player on video, in person, in training, talk to his friends, family, former team mates and coaches... until they have enough information to answer all their questions and concerns.

nytimes.com/2019/05/22/mag…