What impact does a tennis coach have on performance?

Recent years have seen the advent of what have been described as "super coaches". Those of you unfamiliar to the tennis world may expect people who fall into this category to be coaches who get results on a consistent basis - perhaps the tennis equivalent of Antonio Conte, Pep Guardiola or Jose Mourinho in soccer - but this is not actually the case. 

What is a super coach in tennis?

In tennis, a super coach would simply be described as a former top player who is now coach of a top player - in a soccer context, this might be Steven Gerrard or Rio Ferdinand managing the clubs they used to play for.

Assessing player win percentages is useful, albeit inherently flawed in that player injuries - which have no influence from a coach - can affect these numbers.

Without doubt, the comparison between sports is markedly different - anyone who has analysed managerial records in soccer will be able to draw the conclusion that top players do not always make good managers - but it appears that in tennis, having a decent playing career at the very least is virtually a pre-requisite for a coaching job. 

While this pre-requisite can certainly be debated, looking at the numbers behind these ‘super coaches’ could go a long way to proving whether they actually add value, or whether their grandiose nickname is purely hype.

The emergence of the super coach

Since 2010, a number of ‘super coaches’ have taken charge of tennis players. Ivan Lendl (twice) has been Andy Murray’s coach, while Novak Djokovic turned to Boris Becker for several years before moving on to Andre Agassi. Rafa Nadal is now under the tutelage of Carlos Moya following the split with his uncle Toni, while Roger Federer previously employed Stefan Edberg as his coach. 

Even non-elite, but still top players, have employed super coaches. Stan Wawrinka took on Richard Krajicek in 2016, while Marin Cilic was coached by Goran Ivanisevic for several years. The aforementioned Moya and Krajicek have been involved with Milos Raonic, who has a propensity for coach changes - he’s also employed John McEnroe - while Tomas Berdych has also taken on Ivanisevic.

A number of players have certainly been accused of knee-jerk reactions regarding coach recruitment - a facet of tennis which is very relatable to soccer - and it is likely that this process will continue until players are aware of quantifiable metrics which can reliably assess coaching performance.  Much of coaching is still done on an ‘eye test’ or ‘feel’ basis, and this subjective judgement is likely to create bias.

How do you measure the impact a coach has?

Assessing player win percentages is useful, albeit inherently flawed in that player injuries - which have no influence from a coach - can affect these numbers. 

Furthermore, a player’s place on the age curve can have considerable impact, with young players logically able to improve their win percentages, while a veteran may consider maintaining a win percentage from one year to the next to be a relative success. 

There is a case that Ivan Lendl (in his first coaching spell) did not actually improve Andy Murray nearly as much as the media would have you believe.

Arguably the best measure of a coach’s performance is whether they were able to improve their player’s fortunes relative to the betting markets - would backing a player under a given coach yield long-term profits? There would certainly be a case for suggesting that a ‘super coach’ would be able to yield these long-term rewards. 

Therefore, an assessment of the ‘super coaches’ mentioned above was undertaken, for matches in Grand Slams and on the ATP Tour (250 to 1000 level). Davis Cup, where national team coaches are used, were not included in this assessment.

Pinnacle’s closing prices were used, and a hypothetical £100 flat stake bet on each match that a player was coached for by a given coach was adopted as a staking strategy, and this assessment can be seen below (sorted by coach return on investment) :-

*John McEnroe coached Milos Raonic in conjunction with Carlos Moya in McEnroe’s assessed matches 

Evidently, we can see that the vast majority of ‘super coaches’ were unable to add betting value to their players, and thus, it is questionable as to whether they actually improved the level of their players more than generic coaches. 

Of those coaches with a decent sample size of data, only Stefan Edberg, coaching Roger Federer, yielded considerable improvement, while Boris Becker, with Novak Djokovic, did also generate some improvement as well.

However, there is a case that Ivan Lendl (in his first coaching spell) did not actually improve Andy Murray nearly as much as the media would have you believe - variance and small margins play a big part in tennis players’ success and failure at the top level. Richard Krajicek’s data with Milos Raonic, and Goran Ivanisevic’s with Tomas Berdych were both also disappointing.

Overall, super coaches yielded a rather mediocre -0.16% ROI across a sample of around 800 matches, and there is little doubt that - similarly to soccer - there is a reasonable argument to be made that elite players generally do not make for elite coaches.