Michigan Soccer Analytics Society

Student organization @umich dedicated to soccer analytics. Division of @cssmichigan. Check out our other accounts: @analyticsLaLiga, @NUFCAnalytics

[Michigan Soccer Analytics Society @msocanalytics]

Updated the #PremierLeague point share tableau dashboard. Now includes all seasons from 2000/01 to present:

https://t.co/79EKHeqRL4

[Michigan Soccer Analytics Society @msocanalytics]

I made a dashboard in tableau displaying every #PremierLeague player’s point shares for each season since 2011/12. You can search by player, season, and club as well. Will add to it as I pull more seasons.

Check it out here:

https://t.co/ZtDeX2CTjL

[Avi @analyticsavi]

A new way to categorize defense!

I trained an XGBoost model to predict the # of shots a team will block and compared that to their actual blocks to create Blocks Over Expected (BOE)

method: https://t.co/jsvgzAgdKg

interactive viz: https://t.co/9IxbGhxhHf

[Michigan Soccer Analytics Society @msocanalytics]

As I approach graduation, I’m going to start shifting my soccer analytics work to my personal account @mitchelgreen34. Won’t have a ton besides projection models until the end of April, but would appreciate a follow if you want to keep up w the work I’ve been doing here.

[Michigan Soccer Analytics Society @msocanalytics]

@hmmmm123456543 @coachvincent21 End of the day though maybe he’ll do well at Everton. I don’t know. He’s still only 22. Just making some observations that I’ve had watching him week in week out for four seasons

[Michigan Soccer Analytics Society @msocanalytics]

Dwight McNeil was quite literally one of the worst players in the Premier League last year. Burnley just stole £20 million from Everton.

[David Ornstein @David_Ornstein]

🚨 Everton have total agreement with Burnley to sign Dwight McNeil. Fee undisclosed but thought to be ~£20m. 22yo winger at Finch Farm & medical almost complete. If ok, he’ll sign 5yr contract. W/ @adjones_journo @GregOK @TheAthleticUK #EFC #twitterclarets https://t.co/lNEYhivcBV

[Michigan Soccer Analytics Society @msocanalytics]

@hmmmm123456543 @coachvincent21 He also gets caught up to by defenders a lot even when he has tons of space which is probably the bigger problem. You aren’t gonna see that in the numbers though. As far as I know that not something that’s measured

[Michigan Soccer Analytics Society @msocanalytics]

@hmmmm123456543 @coachvincent21 Do you know what the % of his successful attempts was? Feels like based on his play style that could simply be because he tried it so many times (at least that what it seemed from watching him all season)

[Michigan Soccer Analytics Society @msocanalytics]

@hmmmm123456543 @coachvincent21 Add onto that the problem that he can’t beat his man often enough because he isn’t particularly quick and is extremely one footed and he’s going to struggle a ton even with slightly better teammates

[Michigan Soccer Analytics Society @msocanalytics]

@hmmmm123456543 @coachvincent21 Don’t know where you’re getting those numbers but that just isn’t true.

But even if it was, the chances he creates aren’t GOOD chances. That’s the problem. His total creation numbers look okay because he’s getting the ball more than anyone, but the actual quality isn’t there

[Michigan Soccer Analytics Society @msocanalytics]

By the goals created above replacement metric we created, based on his usage rate and goal contributions relative to his position, Maupay sits on 0.01 GCAR per 90.

Sell while you still can, Brighton.

[Michigan Soccer Analytics Society @msocanalytics]

Here’s what our match projection model has for this weekend’s #MLS matches. Have started incorporating betting market projections and 538 odds into our own probabilities to try and catch things our relatively simple model is missing. As such, we are now fairly in line with Vegas.

[Michigan Soccer Analytics Society @msocanalytics]

Gabriel Jesus seems like a good pickup for Arsenal. He’s versatile and can be effective without being an offensive focal point. Curious to see how they decide to use him, but in the past he’s been able to work in multiple different roles for Man City/Brazil so there are options

[Michigan Soccer Analytics Society @msocanalytics]

This will be an ongoing thread of player radars for the biggest transfers of the summer.

Since I only have top 5 data, I’ll only be doing players that moved between teams in top 5 leagues.

Perhaps at some point I’ll cover point shares for Darwin Nuñez, etc. but not for now:

[Michigan Soccer Analytics Society @msocanalytics]

Don’t really know what Barcelona are thinking signing Raphinha. Good player no doubt, but not a great signing when Barca 1) can’t afford him and 2) are way overpaying for him. He should be effective, and it’s good that he’s getting a chance at a higher level, but not great imo.

[Michigan Soccer Analytics Society @msocanalytics]

Here’s what our match projection model has for the midweek MLS matches:

[Michigan Soccer Analytics Society @msocanalytics]

This is a great lesson in framing. The numbers a player puts up only matter relative to the kind of role he has been playing/how you want him to play. Odegaard is an excellent playmaker, but depending on role he can either be a serviceable goal threat or a liability in that area.

[Michigan Soccer Analytics Society @msocanalytics]

Martin Odegaard’s radar is really interesting because of how it changes based on which position group you put him with. With other AM/FWs he shows a weak goal threat but still impressive creation and ball progression and decent ball retention…

[Michigan Soccer Analytics Society @msocanalytics]

But grouped with other midfielders, Odegaard’s creative numbers are off the charts, his shooting numbers now look well above average, but his ball retention numbers are now below avg and his progression numbers are not as impressive.

[Michigan Soccer Analytics Society @msocanalytics]

Haven’t posted model projections in a while but thought I’d bring it back. Here’s our projections for this weekend’s #MLS matches: