Stats and Analysis Thread

[Alan]

Just not from Argentina (the top flight there, not Argentinians full stop) please.

 

[Johnny75]

On video scouting databases do they get like a wide angle/full pitch view these days? I do accept the old 'you don't see everything on TV' side, but wonder if video angles can now give a fuller perspective of the pitch (and indeed a better or at least more consistent view).

I thin it was Mark that mentioned the other day about mental assessments being the next 'big' scouting differentiator, and I do think there's a lot of value in that space. Not just in terms of generally measurable decision making speed etc, but also in how players view the game. Like you can you stats to show how much/often Jota sprints etc, but it can't be log until AI puts a number on the in game intelligence of that pressing, accounting for scanning, looseness of touch, awareness of what lanes are closed off etc. And the same for passing, progression and even movement when in possession. We're at the point where every physical and most technical attribute can be studied and scored, the match intelligence aspect in addition will be a massive boost

Mental state has been looked at for at least 2 decades now.

Can't quantify that, it will always be the eye test.

 

[Del Woppio]

Mental state has been looked at for at least 2 decades now.

Can't quantify that, it will always be the eye test.

Yeah tbf I may be getting a bit carried away with the AI hype. I get that scouting can do the "right, Neves *does* do this and that but having watched him I think he could also do these other things", but also that there would be certain fundamentals or stats that would either identify that talent, or solidify the eye test.

This is probably a really long question rather than a point or an opinion.

Completely made up example, but say you're looking for a 'playmaker', to make progressive passes, assist the assister and chip in himself kind of thing. Now you could look at all the stats that show the right numbers for how you'd expect a playmaker to perform in terms of progression etc. And at this level I think even real life scouts would look at established performers in a position and assessing them at what they actually do in the position of need.

But is there an algorithmic way of accounting for the style of play of a team/manager's instruction when viewing and comparing those numbers. So if you take a midfielder for a team like say Brighton who have higher xG, xT, possession and territory than a team like Palace, and they have a midfielder who's 'playmaker' numbers are lower, but actually as a percentage of his team's numbers are higher than the Brighton player's, do things like the fbref/statsbomb comparison wheels take that into account? So instead of looking at the top progressive passers - which will play for the more attacking managers - can they highlight players that have those skills and abilities but just aren't in the right teams to make the most of those skills.

I haven't read that back and I'm sure it's just a word salad but hopefully the gist is there. I'm not arguing a point either way really just seeing what's done, what's possible and what's fanciful, out of curiosity and boredom. Like a fancy way of finding a player that *could* do X if he was in the right environment

 

[Johnny75]

Yeah tbf I may be getting a bit carried away with the AI hype. I get that scouting can do the "right, Neves *does* do this and that but having watched him I think he could also do these other things", but also that there would be certain fundamentals or stats that would either identify that talent, or solidify the eye test.

This is probably a really long question rather than a point or an opinion.

Completely made up example, but say you're looking for a 'playmaker', to make progressive passes, assist the assister and chip in himself kind of thing. Now you could look at all the stats that show the right numbers for how you'd expect a playmaker to perform in terms of progression etc. And at this level I think even real life scouts would look at established performers in a position and assessing them at what they actually do in the position of need.

But is there an algorithmic way of accounting for the style of play of a team/manager's instruction when viewing and comparing those numbers. So if you take a midfielder for a team like say Brighton who have higher xG, xT, possession and territory than a team like Palace, and they have a midfielder who's 'playmaker' numbers are lower, but actually as a percentage of his team's numbers are higher than the Brighton player's, do things like the fbref/statsbomb comparison wheels take that into account? So instead of looking at the top progressive passers - which will play for the more attacking managers - can they highlight players that have those skills and abilities but just aren't in the right teams to make the most of those skills.

I haven't read that back and I'm sure it's just a word salad but hopefully the gist is there. I'm not arguing a point either way really just seeing what's done, what's possible and what's fanciful, out of curiosity and boredom. Like a fancy way of finding a player that *could* do X if he was in the right environment

I think what you've described is the holy grail of stats and scouting. But it could very well be just a form of modern day alchemy.

Even the great Bill James thinks that stats for fluid sports is next to impossible.

Lots of folks will tell you it's possible but it seems all a bit snake oil to me. And if it were in not sure the game would be better off because of it, if that makes sense.

 

[Deutsch Wolf]

Rather than investing an unspecified amount of money into something so complex with so many variables and nuances for automation, it'd be easier to just watch players. You'll forever have limitations with what software can do.

 

[Aikiwolf]

Ai might have a useful application on body mechanics and identifying trends in players movements that can be mitigated by training or perhaps identifying players who are likely to have an injury prone career.

As a scouting tool it will have uses but statistics don't tell the whole picture of a player but it will help to filter out players outside of certain parameters.

 

[AndyWolves]

I kind of think of stats as the CV and the eye test as the interview.

 

[C.L.D]

As good a place as any. Wondering how well we did out of the 17 teams not relegated/promoted in the '22 calendar year? Not sure how to work that out manually.

 

[Derby Wolf]

As good a place as any. Wondering how well we did out of the 17 teams not relegated/promoted in the '22 calendar year? Not sure how to work that out manually.

It includes relegated sides but 14th of 23.

P:27 W:11 D:6 L:20 F:30 A:55 PTS: 39 GD: -25

Premier League - Annual table

This statistic shows the calendar year table 2022 in the Premier League. It calculates all points won by the teams between January 1st and December 31st.

www.transfermarkt.co.uk

 

[C.L.D]

It includes relegated sides but 14th of 23.

P:27 W:11 D:6 L:20 F:30 A:55 PTS: 39 GD: -25

Premier League - Annual table

This statistic shows the calendar year table 2022 in the Premier League. It calculates all points won by the teams between January 1st and December 31st.

www.transfermarkt.co.uk

A lot better than I thought. Start of year was quite good though.

Thanks!

 

[Lycan]

https://twitter.com/x/status/1694644681596952619

Way too small of a sample size so far but I'd expect this sort of stat to continue if we keep playing the way we are currently and keep those key players. There's a chance here to build a really interesting team with a clear identity, a side which could throw up a different kind of problem for opposition than nearly anyone else in the league.

 

[Dire Wolf]

It's certainly interesting and something to feel encouraged about. We have a few players who are willing to carry the ball at teams and that is going to create opportunities if they continue to be successful.

However, you know which successful dribble doesn't appear on that list? The one by Mitoma that ended in a goal. And ultimately these successful dribbles need to have a successful end result.

 

[Johnny75]

https://twitter.com/x/status/1694644681596952619

Way too small of a sample size so far but I'd expect this sort of stat to continue if we keep playing the way we are currently and keep those key players. There's a chance here to build a really interesting team with a clear identity, a side which could throw up a different kind of problem for opposition than nearly anyone else in the league.

All that ball carrying yet our 1 goal is from a set piece.

It's almost as if it's a bobbins tactic and stat.

 

[prawnking]

... and not a single point on the board to be seen

 

[Lycan]

All that ball carrying yet our 1 goal is from a set piece.

It's almost as if it's a bobbins tactic and stat.

Not sure what you mean?

A team that can set up to be compact and also have multiple players with the ability to carry the ball like this isn't a problem opposition teams face too often.

That's the point I was making, that there's a chance to build a team with a unique identity.

 

[Johnny75]

Not sure what you mean?

A team that can set up to be compact and also have multiple players with the ability to carry the ball like this isn't a problem opposition teams face too often.

That's the point I was making, that there's a chance to build a team with a unique identity.

I understand your point and it's certainly an interesting concept to have a team of successful ball carriers as a tactic.

One I don't think we've seen since the 1930's.

If we're going to do that though we need those ball carriers to make better decisions with their end products. Currently we have great ball carriers with zero end product. That isn't sustainable.

 

[YoungWolf]

Not sure what you mean?

A team that can set up to be compact and also have multiple players with the ability to carry the ball like this isn't a problem opposition teams face too often.

That's the point I was making, that there's a chance to build a team with a unique identity.

We’ve also created quite a lot of chances (certainly compared to usual anyway!)

 

[YoungWolf]

Not watched it myself but may be interesting

 

[Deutsch Wolf]

We cor shoot, Franksie.

 

[C.L.D]

Black on black, their graphic designer deserves a roasting.