Economist article on Manager Impact in Pro Football

[Midas Mulligan]

I’m intrigued by this piece I saw on Reddit.

Some of you geeks take a look and let me know what you think. I don’t want to shape discussion but will weigh in later.

(Couldn’t resist a little Easter egg for dummyrun)

 

[Ulrich]

I’m intrigued by this piece I saw on Reddit.

Some of you geeks take a look and let me know what you think. I don’t want to shape discussion but will weigh in later.

(Couldn’t resist a little Easter egg for dummyrun)
View attachment 9456

Interesting ideas, but I think it is missing a key component to tie things together - the checking of the manager's ego at the door and playing to the strengths of the players.

Case in point.... Dome arrived and won his first set of matches by continuing with the winning tactics Vieiera had instilled and utilizing the players in roles they were familiar with and maximized their skill sets. Once Dome made his mark on the team by forcing change through endless tinkering and utilization of players in unfamiliar roles, the team looked like a disjointed trash heap. He brought in a player he was familiar with to start games while only having a fraction of the ability compared to current players. Dome refused to use his round pegs in round holes.

The flip side of this is ManU bringing in Solskjær who immediately changed the team's defensive & losing tactics to suit the players' skill sets which translated to fast/attacking forays. He's been rewarded with 7 wins/0 losses across all competitions, his players show energy and enthusiasm that was missing under previous manager, and players that had been marginalized have miraculously demonstrated complete competency holding down their roles (Phil Jones as solid CB, Rashford is hottest CF after only being played as a wing previously, Pogba is a linchpin monster again). All this after not bringing in a single new player and strictly using the squad that had so badly performed under Jose.

Maybe the difference is that Dome was handed the role and didn't have to prove success to keep the role so he fcked around and tinkered instead of producing positive results. On the flip side, Solskjær was named interim with little chance of being appointed full-manager, and he's in turn delivered results that beg to include him in the real selection process. Perhaps Solskjær's success will turn downward, but that's unlikely since he's already made his changes that are why his dividends are being paid - each match his squad looks more and more confident with what they're doing.

 

[ManhattanValley]

Simon Kuper has done some excellent work on this already (I'm sure I'm preaching to the converted, but Soccernomics and Soccer Men are both fantastic reads)

He makes the argument that the narrative of managers as tactical and/or motivational geniuses is mostly irrelevant compared to the wage bill, and that smart recruitment and data analysis is what can make a difference.

https://www.ft.com/content/0f79860e-605e-11e6-ae3f-77baadeb1c93 (paywall, but worth the click)

Plus, his argument on the honeymoon period for new managers has made me forever skeptical about a new manager's results after the previous guy was sacked.

https://www.fourfourtwo.com/us/features/soccernomics-does-sacking-manager-actually-make-difference

 

[Seth]

Interesting ideas, but I think it is missing a key component to tie things together - the checking of the manager's ego at the door and playing to the strengths of the players.

Case in point.... Dome arrived and won his first set of matches by continuing with the winning tactics Vieiera had instilled and utilizing the players in roles they were familiar with and maximized their skill sets. Once Dome made his mark on the team by forcing change through endless tinkering and utilization of players in unfamiliar roles, the team looked like a disjointed trash heap. He brought in a player he was familiar with to start games while only having a fraction of the ability compared to current players. Dome refused to use his round pegs in round holes.

The flip side of this is ManU bringing in Solskjær who immediately changed the team's defensive & losing tactics to suit the players' skill sets which translated to fast/attacking forays. He's been rewarded with 7 wins/0 losses across all competitions, his players show energy and enthusiasm that was missing under previous manager, and players that had been marginalized have miraculously demonstrated complete competency holding down their roles (Phil Jones as solid CB, Rashford is hottest CF after only being played as a wing previously, Pogba is a linchpin monster again). All this after not bringing in a single new player and strictly using the squad that had so badly performed under Jose.

Maybe the difference is that Dome was handed the role and didn't have to prove success to keep the role so he fcked around and tinkered instead of producing positive results. On the flip side, Solskjær was named interim with little chance of being appointed full-manager, and he's in turn delivered results that beg to include him in the real selection process. Perhaps Solskjær's success will turn downward, but that's unlikely since he's already made his changes that are why his dividends are being paid - each match his squad looks more and more confident with what they're doing.

With sincere apologies:

Interesting ideas, but I think it is missing a key component to tie things together - the checking of the manager's ego at the door and playing to the strengths of the players.

Case in point.... Solskjær arrived and won his first set of matches by calming the more or less open player revolt. Whether the players were right or wrong doesn't matter as perhaps anyone United brought in might have gotten the same reaction.

The flip side of this is us bringing in Domé who took over a team that was doing well more or less under a well-liked manager who was also a proven player. Taking over a team halfway through a season is almost never a good thing, especially when they're doing reasonably well.

Maybe the difference is that Dome was handed the role and didn't have to prove success to keep the role, at least for this season. If we're having these same conversations at the end of the new season however, well, then I'll yield that you have a valid point.

Bottom line for me is that it's still only the middle of January and we don't even have a full lineup yet. If they bring in a couple of new amazing players and we end up having a great season then all this Domé-out talk fades away. If we're still at this point in a year, well, perhaps I'll start sharpening my pitchfork along with all o' youse.

 

[ManhattanValley]

Case in point.... Solskjær arrived and won his first set of matches by calming the more or less open player revolt. Whether the players were right or wrong doesn't matter as perhaps anyone United brought in might have gotten the same reaction.

The flip side of this is us bringing in Domé who took over a team that was doing well more or less under a well-liked manager who was also a proven player. Taking over a team halfway through a season is almost never a good thing, especially when they're doing reasonably well.

As per that second link I posted up there, it shouldn't be surprising that a new manager will more likely outperform his predecessor when that team is playing badly (2019 Man Utd) than when the team is playing really well (2018 NYCFC).

Like you say, it's too early to say what's up with Dome. But I wouldn't be any less surprised if he turns out to be pretty good than I would be to realise that OGS isn't suddenly a managerial savant.

 

[Ulrich]

With sincere apologies:

Interesting ideas, but I think it is missing a key component to tie things together - the checking of the manager's ego at the door and playing to the strengths of the players.

Case in point.... Solskjær arrived and won his first set of matches by calming the more or less open player revolt. Whether the players were right or wrong doesn't matter as perhaps anyone United brought in might have gotten the same reaction.

The flip side of this is us bringing in Domé who took over a team that was doing well more or less under a well-liked manager who was also a proven player. Taking over a team halfway through a season is almost never a good thing, especially when they're doing reasonably well.

Maybe the difference is that Dome was handed the role and didn't have to prove success to keep the role, at least for this season. If we're having these same conversations at the end of the new season however, well, then I'll yield that you have a valid point.

Bottom line for me is that it's still only the middle of January and we don't even have a full lineup yet. If they bring in a couple of new amazing players and we end up having a great season then all this Domé-out talk fades away. If we're still at this point in a year, well, perhaps I'll start sharpening my pitchfork along with all o' youse.

You kinda prove my point. Calming the players at ManU by playing to their strengths and not forcing tactics on them that are solely based on a Managers decision is absolutely the act of checking one’s ego. Solskjær didnt decide on tactics in a vacuum without considering what could be accomplished/accepted by the players (as Jose did), he managed to the skillset of his players. Whether another manager could have done the same is a possibility, but that is also an unknown. ManU is now finding success using power/pace which is what the players are comfortable with and also happen to be what the club is known for.

Dome had a perfect situation set on a plate for him and the results he got the first set of matches showed so with the wins by using PV’s tactics. I’m not sure what is bad about that situation that you allude to - it was easy to keep the status quo and there was zero reason to tinker (unlike at ManU where the team was sliding miserably). Dome took it upon himself to move away from the formation and assignments that had brought success when it wasn’t required to quell an open revolt or to look for a spark to turn around the team’s fortunes. It was an arbitrary decision predicated on ego - even with injuries, team tactics could be maintained as that’s what the club (including subs) had trained for all season - why have backups that can’t play the same role of the starter?

Planets will have to realign for me to stray from DomeOut. The closest comparison I can think of for him is he’s the equivalent of Avram Grant without the success Grant initially had at Chelsea (by continuing to use Jose’s tactics).

 

[Rimil]

I am not a trained statistician but this just seems like pseudo-science. I appreciate the attempt but what observable measure is this study even grounded in? They are using pre-season subjective player stats from a video game as a constant to test against. Its a one time look. Doesn't adjust for player stat improvements/injuries and/or intra-season transfers it seems. To justify using the video game stats they fit known information (i.e. historical standings) over the stats to "prove" its a decent measure. So they say. I don't actually know if it is meaningful. They say it comes withing 8pts +/- of actual results. Then they go on to say mourinho outperforms slightly with +1 over the average?! Using a scale that had 8pts of error they say he outperforms the average by 1? I don't know man; sketchy science.

I could just as easily comb through millions of bits of information to find what has a better fit. Then write an article around it. (hypothetical!) "Air pressure over a team's third preseason match more important than the players" That sounds crazy right, but I can show you a model that proves over 20,000 matches, if you played your third match in exactly 29.80mb you will outperform your expected standings as proven by one static look at pre-season fifa ratings. Its called P-value hunting and its unethical in stats.

But anyway, they used a modified version of p-value hunting to feel confident these fifa stats are significant (even though I still think that margin of error is ridiculous). I say modified, because instead of scraping data for anything and getting something that sounds stupid like air pressure over a single pre-season match which might statistically yeild a better p-value, they just tested for managers. They must have felt it was significant enough p-value to go ahead and write the article.

Obviously sounds more credible to write an article comparing a manager to a teams performance than air pressure, but then you look at the measure and its grounded is silly subjective data that they prove is valid because it fits results with a margin of error I could drive a tractor trailer through. So doesn't that seem a little crazy to prove a manager's worth off of static fifa ratings?

Good managers are naturally going to fall on the positive side of test results. When you try to normalize the scale to see any deeper meaning like outperforming expected results it matters how you define expected results. I don't feel like this article really used a credible source. Makes for fun debate, but I don't glean any true scientific facts from it other then they found a scale that poo-poo's ancelotti, and praises simeone/klopp and comes down so-so on mourinho. Feels like that data-mined for general sentiment results and passed it off as statistics.

 

[Christopher Jee]

I am not a trained statistician but this just seems like pseudo-science. I appreciate the attempt but what observable measure is this study even grounded in? They are using pre-season subjective player stats from a video game as a constant to test against. Its a one time look. Doesn't adjust for player stat improvements/injuries and/or intra-season transfers it seems. To justify using the video game stats they fit known information (i.e. historical standings) over the stats to "prove" its a decent measure. So they say. I don't actually know if it is meaningful. They say it comes withing 8pts +/- of actual results. Then they go on to say mourinho outperforms slightly with +1 over the average?! Using a scale that had 8pts of error they say he outperforms the average by 1? I don't know man; sketchy science.

I could just as easily comb through millions of bits of information to find what has a better fit. Then write an article around it. (hypothetical!) "Air pressure over a team's third preseason match more important than the players" That sounds crazy right, but I can show you a model that proves over 20,000 matches, if you played your third match in exactly 29.80mb you will outperform your expected standings as proven by one static look at pre-season fifa ratings. Its called P-value hunting and its unethical in stats.

But anyway, they used a modified version of p-value hunting to feel confident these fifa stats are significant (even though I still think that margin of error is ridiculous). I say modified, because instead of scraping data for anything and getting something that sounds stupid like air pressure over a single pre-season match which might statistically yeild a better p-value, they just tested for managers. They must have felt it was significant enough p-value to go ahead and write the article.

Obviously sounds more credible to write an article comparing a manager to a teams performance than air pressure, but then you look at the measure and its grounded is silly subjective data that they prove is valid because it fits results with a margin of error I could drive a tractor trailer through. So doesn't that seem a little crazy to prove a manager's worth off of static fifa ratings?

Good managers are naturally going to fall on the positive side of test results. When you try to normalize the scale to see any deeper meaning like outperforming expected results it matters how you define expected results. I don't feel like this article really used a credible source. Makes for fun debate, but I don't glean any true scientific facts from it other then they found a scale that poo-poo's ancelotti, and praises simeone/klopp and comes down so-so on mourinho. Feels like that data-mined for general sentiment results and passed it off as statistics.

This seems valid.

I also question whether those FIFA ratings are a good source of foundational data due to the fact that they are difficult to isolate from manager influence. If I'm not mistaken, you could conclude that the FIFA ratings are more tightly correlated with manager influence than player influence. You could probably also conclude from this article that the impact of players like Messi and Ronaldo aren't adequately represented by their FIFA ratings.

 

[Midas Mulligan]

I am not a trained statistician but this just seems like pseudo-science. I appreciate the attempt but what observable measure is this study even grounded in? They are using pre-season subjective player stats from a video game as a constant to test against. Its a one time look. Doesn't adjust for player stat improvements/injuries and/or intra-season transfers it seems. To justify using the video game stats they fit known information (i.e. historical standings) over the stats to "prove" its a decent measure. So they say. I don't actually know if it is meaningful. They say it comes withing 8pts +/- of actual results. Then they go on to say mourinho outperforms slightly with +1 over the average?! Using a scale that had 8pts of error they say he outperforms the average by 1? I don't know man; sketchy science.

I could just as easily comb through millions of bits of information to find what has a better fit. Then write an article around it. (hypothetical!) "Air pressure over a team's third preseason match more important than the players" That sounds crazy right, but I can show you a model that proves over 20,000 matches, if you played your third match in exactly 29.80mb you will outperform your expected standings as proven by one static look at pre-season fifa ratings. Its called P-value hunting and its unethical in stats.

But anyway, they used a modified version of p-value hunting to feel confident these fifa stats are significant (even though I still think that margin of error is ridiculous). I say modified, because instead of scraping data for anything and getting something that sounds stupid like air pressure over a single pre-season match which might statistically yeild a better p-value, they just tested for managers. They must have felt it was significant enough p-value to go ahead and write the article.

Obviously sounds more credible to write an article comparing a manager to a teams performance than air pressure, but then you look at the measure and its grounded is silly subjective data that they prove is valid because it fits results with a margin of error I could drive a tractor trailer through. So doesn't that seem a little crazy to prove a manager's worth off of static fifa ratings?

Good managers are naturally going to fall on the positive side of test results. When you try to normalize the scale to see any deeper meaning like outperforming expected results it matters how you define expected results. I don't feel like this article really used a credible source. Makes for fun debate, but I don't glean any true scientific facts from it other then they found a scale that poo-poo's ancelotti, and praises simeone/klopp and comes down so-so on mourinho. Feels like that data-mined for general sentiment results and passed it off as statistics.

Doesn't the article say they could predict the final league table within an error margin of 8 points? Is that the same as saying they could predict a team's results within that same margin of error? I wouldn't read it that way, as I'm pretty sure I could just project each teams' point totals within 8 without any data whatsoever.

This seems valid.

I also question whether those FIFA ratings are a good source of foundational data due to the fact that they are difficult to isolate from manager influence. If I'm not mistaken, you could conclude that the FIFA ratings are more tightly correlated with manager influence than player influence. You could probably also conclude from this article that the impact of players like Messi and Ronaldo aren't adequately represented by their FIFA ratings.

This is where I have a problem. If I'm pointing out anything, it's that a manager actually hurts his own ability to outperform if he improves the abilities of his players year over year.

Let's look at Gini Wijnaldum - sort of middle of the road to above-average winger/10 on a terrible team. Moves to Liverpool, is deployed with different instructions and different positions. 3 years later, he's among the best holding mids in the world. How do you isolate that from the manager's impact? I don't think you can.

Then again, no one claims this method is perfect. Like any real world instance, the researcher is seriously limited in his ability to control for all of the variables.

I mean, we haven't even touched on the manager's impact in recruitment. Can Klopp get better players than Brodge? Hell yes. And as a result, he gets his first choices more often than many other managers. Virtuous cycle at its finest. I suppose they are trying to hone in only on tactics and in-season, on-field sort of stuff.

I don't have any conclusions. I didn't bring this here as though it were handed to me from God as another testament. It's just sort of interesting.

I actually found the idea that you could predict a league table within 8 points using FIFA ratings pretty freaking cool. I find that to be far better than I'd like to credit EA.

 

[dummyrun]

Doesn't the article say they could predict the final league table within an error margin of 8 points? Is that the same as saying they could predict a team's results within that same margin of error?

Yup. Eight points of mean absolute error per team actually isn't bad: it's about equal to FiveThirtyEight's average error in projecting the MLS table the last couple years (MLS is harder to predict than top European leagues, but Nate Silver's model is a lot better than FIFA player ratings). If the average error here were eight points for a whole league, they wouldn't be telling us about it, they'd be out making jillions off spread betting.

Rimil is right that the quick and dirty methodology makes narratives about individual players or coaches suspect. But it might still support the top-line conclusion about the impact of managers as a population.