Green means go...Away to Lincoln , Mon 21st Apr. 3-0'clock.

[Prufrock]

Am currently reading Ian Graham's "How to Win the Premier League" which is interesting for two reasons, 1) on data analysis in football generally, and 2) given he heads Ludonautics who I think now have equity in the club, and who you'd imagine are involved in our data side.

Couple of reflections:

1) The lines of accountability one is definitely tricky. He talks about the differences in how it worked under Brendan Rogers who didn't buy in and Klopp who did. The idea seems to be that they would feed into the scouting department in e.g. drawing up shortlists, spotting good players scouting didn't like (e.g. Joel Matip, looks gangly, made a few high profile mistakes but actually v good player on a free). And under Klopp all departments had a veto, they didn't sign anyone data said was a poor value signing, but equally if the manager didn't want them, or the scouts didn't, they weren't signed. Ultimately it sounds like it was the manager held responsible. Rogers didn't want to engage or hear their point of view, and was able to sign e.g. Benteke against their recommendation. Scouting dept picked Markovic against their recco, but then under Klopp they had a tick. They persuaded him to sign Salah over Brandt based on data.

He's obviously not impartial, but he's also fairly open about what didn't work, and doesn't claim credit where not due (e.g. Van Dijk, Alisson, the data said they were excellent but scouting and the manager already knew that. Firmino they recommended but thought was over-priced at what they paid. On the other hand Salah, Mane, Wijnaldum, Robertson two big wins.

2) How the data works - interesting and validating to read xG is basically bollocks other than long term trends and trying to work out where luck enters things. They have lots of much more complicated tools for analysis, which are position specific. Their particular "thing" is possession value added, which basically tries to work out who has good stats for safe passing but doesn't add much (e.g. Joe Allen) and who gives it away more but adds value (Gerrard or De Bruyne). Obviously you don't need data to tell you the latter two are better than the former, but it helps differentiate.

3) How we might use it - one of the interesting things was how the models can now recreate pretty good analysis that used to require e.g. prozone, just using video analysis. I do wonder if the recent TV deals with all lower league games being filmed are why they have got involved. And that we'll be using that to try to find the best players in e.g. L2 with more info. I also imagine SS and Harkin are recruited to fit into this system. Pretty clear the manager remains the most important person (unless you have a footballing DoF and head coach only) and it seems Klopp's willingness to work with them and discuss and compromise was the reason it worked.

Also he talks a lot about how a big part of their success was recognising when not to make signings. Their aim wasn't to find the best players, but to find the best value players. Someone like James Milner, free, can play several positions was vital in what they did, as was the money they got in for Coutinho. But roughly 50% of all transfers fail, and the data is there as an added layer of due diligence to try to a) increase that % but b) as importantly let you recognise where a shiny new signing is no better than what you already have.

That is clearly going to be a tricky line for us to tread this summer.

[BWFC_Insane]

As one of "some who bristle," I have to say I'm not convinced by this apologising crap as some sort of mass group therapy, if they do, they do. We've been shit compared to what we should've expected this season, a few people are now seeking their next career moves because of it.

I can understand why someone who missed out by 4? points and / or one game at Wembley isn't maybe an instant dismissal.

I doubt Sharon, or any Board Chairmen don't take counsel before making big decisions, whether that's from Manager, DoF, their other Board Members or elsewhere. Probably why it's referred to on MotD as "the Emergency Board metting going on behind us."

There have been plenty of "big teams," stuck in L1 for longer than they'd like, switching managers, chucking cash at it.

I think I'm more worried about who's advising, the Chairman's main job is to ensure we got some money to do shit with. Others spend it.

Who they take advice from is clearly very important but given someone or a group of people make the decisions and given Evatt and not dismissing him on several occasions was a huge mistake I think it’s obvious that those who make those decisions have to accept the criticism for them. Sharon may not be hands on. Maybe the board overruled her. Maybe she got bad advice. Maybe it was all Markham’s choice. I don’t know - but that ultimately is where accountability lies. Wherever those decisions are made. A manager can be hired and dreadful. The decision to hire them then has to be scrutinised. Cos the manager will waltz off eventually. Same with player recruitment. There needs to be accountability with the DoF or head coach or whoever it is that makes the decision. Because as much as we get annoyed with players they are only here because someone paid them and in some cases their previous club a lot of money.

FWIW I don’t think we can only criticise Evatt or Markham. Those who left them there have to take their portion of blame. And in the future too. In the same way Gartside took the credit for Allardyce and the blame for what happened after. It’s all part of it. Building the right team etc…

[Worthy4England]

A good read, Pru, thanks. Given they were here in summer, did it have a Chapter on "How to count your number 10's?"

[Worthy4England]

As one of "some who bristle," I have to say I'm not convinced by this apologising crap as some sort of mass group therapy, if they do, they do. We've been shit compared to what we should've expected this season, a few people are now seeking their next career moves because of it.

I can understand why someone who missed out by 4? points and / or one game at Wembley isn't maybe an instant dismissal.

I doubt Sharon, or any Board Chairmen don't take counsel before making big decisions, whether that's from Manager, DoF, their other Board Members or elsewhere. Probably why it's referred to on MotD as "the Emergency Board metting going on behind us."

There have been plenty of "big teams," stuck in L1 for longer than they'd like, switching managers, chucking cash at it.

I think I'm more worried about who's advising, the Chairman's main job is to ensure we got some money to do shit with. Others spend it.

Who they take advice from is clearly very important but given someone or a group of people make the decisions and given Evatt and not dismissing him on several occasions was a huge mistake I think it’s obvious that those who make those decisions have to accept the criticism for them. Sharon may not be hands on. Maybe the board overruled her. Maybe she got bad advice. Maybe it was all Markham’s choice. I don’t know - but that ultimately is where accountability lies. Wherever those decisions are made. A manager can be hired and dreadful. The decision to hire them then has to be scrutinised. Cos the manager will waltz off eventually. Same with player recruitment. There needs to be accountability with the DoF or head coach or whoever it is that makes the decision. Because as much as we get annoyed with players they are only here because someone paid them and in some cases their previous club a lot of money.

FWIW I don’t think we can only criticise Evatt or Markham. Those who left them there have to take their portion of blame. And in the future too. In the same way Gartside took the credit for Allardyce and the blame for what happened after. It’s all part of it. Building the right team etc…

I don't think Sharon was "blameless." There was verbal backing for Evatt in early June, then whispers to BEN after Huddersfield in Sept that he still had backing. You might in a generous mood say "having backed him over Summer, then had to give it more time," as some on here suggested.

By the time we get to 18th? Dec and more verbal backing, I think pretty much everyone was shaking their heads. Then in rolls another 10....

[Dave Sutton's barnet]

Am currently reading Ian Graham's "How to Win the Premier League" which is interesting for two reasons, 1) on data analysis in football generally, and 2) given he heads Ludonautics who I think now have equity in the club, and who you'd imagine are involved in our data side.

Cheers Pru - fascinating stuff. (Although I hope they don't actually increase to 50% failure rate.... )

Feels like a post you'll want to come back to and won't be able to find it 'cos you put it in a match thread....

[Prufrock]

A good read, Pru, thanks. Given they were here in summer, did it have a Chapter on "How to count your number 10's?"

Ha! Pretty clear in the model that the data department (the "research department") had no say on the tactical side of recruitment. If the manager said he wanted 3 number 10s, they'd pull up a shortlist to try to find them.

Interesting that Klopp was very clear he preferred players with different making capabilities. I.e. the data said Robertson wasn't great defensively but Klopp was happy with that as he wanted attacking threat and it was his job to mitigate that weakness. Also open that that turned out to be wrong anyway and Robertson is excellent defensively.

He puts that down to the models not being great at predicting defensive ability at that time. They would pick out the e.g. a goal was scored from a cross from your left, and it would auto attribute blame to the left back. But it wouldn't tell you if it was really the LB's fault, it whether they'd done what they could buta centre half for under the ball, or they'd had to drop in to cover for someone out of position. They worked really hard to build a bespoke model for defenders and seemed happy they'd got there. Hopefully we have access to that this summer!

[Prufrock]

Am currently reading Ian Graham's "How to Win the Premier League" which is interesting for two reasons, 1) on data analysis in football generally, and 2) given he heads Ludonautics who I think now have equity in the club, and who you'd imagine are involved in our data side.

Cheers Pru - fascinating stuff. (Although I hope they don't actually increase to 50% failure rate.... )

Feels like a post you'll want to come back to and won't be able to find it 'cos you put it in a match thread....

Doh X2.

Hopefully no-one else uses the word Ludonautics again and I can find it that way

[truewhite15]

Am currently reading Ian Graham's "How to Win the Premier League" which is interesting for two reasons, 1) on data analysis in football generally, and 2) given he heads Ludonautics who I think now have equity in the club, and who you'd imagine are involved in our data side.

Cheers Pru - fascinating stuff. (Although I hope they don't actually increase to 50% failure rate.... )

Feels like a post you'll want to come back to and won't be able to find it 'cos you put it in a match thread....

Doh X2.

Hopefully no-one else uses the word Ludonautics again and I can find it that way

Ludonautics.
Ludonautics.
Ludonautics.

[Worthy4England]

A good read, Pru, thanks. Given they were here in summer, did it have a Chapter on "How to count your number 10's?"

Ha! Pretty clear in the model that the data department (the "research department") had no say on the tactical side of recruitment. If the manager said he wanted 3 number 10s, they'd pull up a shortlist to try to find them.

Interesting that Klopp was very clear he preferred players with different making capabilities. I.e. the data said Robertson wasn't great defensively but Klopp was happy with that as he wanted attacking threat and it was his job to mitigate that weakness. Also open that that turned out to be wrong anyway and Robertson is excellent defensively.

He puts that down to the models not being great at predicting defensive ability at that time. They would pick out the e.g. a goal was scored from a cross from your left, and it would auto attribute blame to the left back. But it wouldn't tell you if it was really the LB's fault, it whether they'd done what they could buta centre half for under the ball, or they'd had to drop in to cover for someone out of position. They worked really hard to build a bespoke model for defenders and seemed happy they'd got there. Hopefully we have access to that this summer!

A cursory glance at what unfolds in front of us, should be everything we needed to see, to know the models weren't great at predicting defensive ability! I've been saying that for at least three years!

We see it in my business lots. Yards of data poorly interpreted that has some pretty key omissions - sure the models will continue to get better at what they're doing.

Fcking xG my arse!

[Prufrock]

Aye for sure. What you want and need is people who are able to smell test the results to work out where they are deficient.

Two examples - one is that possession value added metric. Brendan Rogers wanted Joe Allen partly because he wanted to dominate the ball, and he had an exceptional pass completion rate. But their modelling showed that was because he made lots and lots of safe passes that didn't increase the teams chances of scoring. What they used instead was a metric that showed how much a player increased chances of scoring with passes (minus the chance of scoring lost and chance of scoring gained when they gave it away). They built that as a result of realising that flaw.

2) for keepers, they took xG (useless other than long term trends for overall performances) and built what they call post-strike xG. I.e. you can now value the actual shot, how well taken it is etc. This allows you to find better strikers (higher PSxG than xG) and goalkeepers (lower goals conceded than PSxG). To take Iles' example of our keepers having low shot saved percentages. Is that because they're not very good, or is it because useless defenders keep giving the other team golden chances? Or are they unlucky against well taken goals?

[Prufrock]

Am currently reading Ian Graham's "How to Win the Premier League" which is interesting for two reasons, 1) on data analysis in football generally, and 2) given he heads Ludonautics who I think now have equity in the club, and who you'd imagine are involved in our data side.

Cheers Pru - fascinating stuff. (Although I hope they don't actually increase to 50% failure rate.... )

Feels like a post you'll want to come back to and won't be able to find it 'cos you put it in a match thread....

Doh X2.

Hopefully no-one else uses the word Ludonautics again and I can find it that way

Ludonautics.
Ludonautics.
Ludonautics.

as long as you keep it in this thread.