The inevitable discourse around the future of a few Premier League managers has begun in recent weeks. Thanks to Tottenham’s stuttering start (to be discussed later), Arsenal’s loss to a good (better than most will give credit for) Sheffield United side, and Man Utd doing their best to emulate Harchester United, it’s safe to say the English papers certainly aren’t starved for subject matters, now Raheem Sterling is getting the public backing he’s deserved his whole career.
With some fans, journalists and TV pundits calling for clubs to swing the reactionary axe, perhaps it’s time to consider the way we evaluate manager’s success. For what it’s worth, I’d assume there’s a pretty strong correlation between the ones who believed Solskjær was the new Alex Ferguson back in March and who now believe Woodward has to hire a Director of Football.
The benefit of hindsight is a wonderful thing, but it must be said that general consensus amongst the football analytics community was that Man Utd’s run of form under Solskjær as the Interim Head Coach was the perfect storm of an relatively easy run of games, over performance (aka luck), and the ‘breath of fresh air’ a new manager often brings before performances tend to regress to the mean. Great news for Ole, bad news for Man Utd. It seems at the time, the only people concerned with the appointment were the spreadsheet nerds. Shortly after the analysts shared their views, the journalists who hate them inevitably shared theirs.
The Man Utd case study is one of many, where looking just a little deeper than wins and losses, and taking emotional bias out of decision making in an industry worth billions, might perhaps be a worthwhile exercise. Amazing revelation!
In short, a well-run club with some clear principles and decision-making processes in place may have made a different decision.
It’s important to stress that even some of the best run clubs in the World in the same circumstance may still have made the same decision, and to blindly blame Ed Woodward for hiring Ole full-time would be naïve. Man Utd’s short-term purple patch backed him into a very difficult corner.
Speaking of well-run clubs, one of the arguments I’ve seen this week surrounding Tottenham, is that Pochettino hasn’t won them anything, and doesn’t have ‘what it takes’ to take them to the next level. Amazing really when you consider they were in the Champions League final only a few months ago.
Consider this however: in 2016/2017 Tottenham were the best team in England over a 38 game rolling period and would likely have won the Premier League. Unfortunately for Spurs, their league winning form was split between two seasons, meaning they never won a title, despite being the best team in the country.
If you put the 2017 Premier League trophy on his list of achievements at Spurs, I believe there would be far fewer pundits and fans calling for a change at the top.
Luckily, Tottenham are a smart club with some excellent decision-making processes in place. They know when and why to hire and fire a manager, and certainly have a model in place that compensates for underlying metrics, rather than wins and losses.
This isn’t to say that Pochettino won’t be fired (although I don’t think this is even a conversation at Spurs currently) rather that any decision made will be a lot more purposeful and targeted than many people believe it should be.
Football Intelligence agency, 21st Club have previously pitched the idea of evaluating Head Coaches differently. Instead of simply looking at results, should football clubs look at different metrics in combination or even use an algorithm to evaluate a manager’s performance?
Interestingly, in Major League Soccer, it was recently reported that former Real Salt Lake coach Mike Petke had bonuses in his contract for playing academy graduates. A case could be made that this becomes a conflict of interest when faced with personnel decisions on match day, but isn’t that exactly what such a clause is designed to do?
If a football club values it’s academy, and believes that playing these players is key to their long-term success, then they probably should be pressuring the Head Coach to play academy graduates, or certainly making the thought of playing them more appealing vs the perceived risk involved.
Believe it or not, some football club owners would like to make money from their investment, rather than throwing their personal fortune into the ever-expanding black hole of in football. With the terrible recent situations at Bury and Bolton, plus the growing list of clubs selling stadiums to themselves and avoid financial fair play penalties, managers who develop young players to sell-on may soon see their stick rise.
One such group who does things rather well is the Red Bull’s. Lead by Sporting Director Ralph Rangnick, they implement a decision-making process geared towards succession planning and continuity.
I recently wrote about why they hired new Leipzig head coach, Jesse Marsch here. Using their high tempo, high pressing style of play as a template, it’s likely that their playing model makes up a large part of manager performance evaluation. Some example key performance indicators being used may be:
- 25% of all available minutes played by U23 players
- PPDA score x or better across a pre-defined period of games or rolling average
- 50% of clear cut chances produced within 5 passes of a turnover
- Rolling average of expected goal difference
(PPDA is a generic metric, designed to measure pressing intensity)
Using these 4 example indicators in conjunction, giving each their own weighting depending on their perceived importance, a club could feasibly use an algorithm to evaluate manager performance. Common sense and a human element should always play a role here, as it should withal analytic integration into decision-making.
A few scenarios to consider:
First team isn’t winning games:
Our expected goal difference is fine, and we know this will lead to positive results in the long term.
Expected goal difference is a concern – we think we will begin to lose games soon
This is something we need to keep an eye on and improve, but academy graduates and young players are featuring a lot. This will lead to profits in the transfer market, which can be reinvested in infrastructure or new players. How is our PPDA? Are we pressing and playing in a way, which represents our club philosophy?
Underlying numbers are poor, PPDA is not great and young players aren’t getting minutes
Fire up the coffee machine, it’s time for an important meeting!
This is a simplified version of how a true managerial evaluation type algorithm would work, and yes, there’s definitely some subjective considerations, which still need to take place. You may even argue that these are similar questions a board would ask (whether deliberately or not) regardless of an algorithm.
- What if this information was made public before a manager signed his contract?
- What if the club came out and named 3 priorities, such as player development, investment in infrastructure and a defined playing style?
- What if after each game a club completely opened up about some basic KPI’s for fans to track?
- What if when the journalists begin to write their articles, the club is already transparent about expectations and regularly communicating the results?
- In the Man Utd case of underlying expected goals, would they have supported Ole getting the job given publicly available numbers?
- What if pundits and Spurs fans knew that they were good enough over 365 days to win the league a couple of seasons ago, and had just reached a Champions League final?
- Given what we know about Man Utd, could a similar process have resulted in a different outcome and prevented another rebuilding period?
Ole is most certainly at the wheel, but how good does it actually feel?