# Should We Be Surprised by Murray's 2017 Clay Performance So Far?

At this point in the clay season, many of us were expecting Andy Murray to have appeared in at least one or more finals and maybe even snagged a title. But with a Round of 16 loss this week to Borna Coric at the Mutua Madrid Open, Murray has crashed out earlier than expected for his third clay court event of the season.

By all accounts Murray’s clay results in 2017 have been disappointing. But, given that even the best players have some chance of a loss, perhaps what we are seeing isn’t anything more than an unlucky streak.

How troubled should we be by Murray’s clay losses all comes down to expectations. Going into 2017, for example, we might have expected him to perform as well or better than 2016. At this stage in the calendar in 2016, Murray had played only two events compared to three events this year, when he added Barcelona to his schedule. Still, by the end of the Madrid Masters in 2016 he had won nearly twice the matches he has won so far during the 2017 clay court swing (7 versus 4).

But even this much of a drop in 2017 could be explained by tougher draws. If Murray has faced more difficult opponents, we would anticipate some extra loses.

*So how does the strength of his opponents on clay in 2016 compare to 2017?*

One way we can factor in opponent difficulty is by comparing how Murray performed in his clay matches against how he was *expected* to perform. Our expectations are closely tied to the opponent’s ability. These expectations tell us for example, that we should be less surprised by a loss to Rafael Nadal than a loss to Albert Ramos-Vinolas. In this post, the exact calculation of what we “expect” comes from the Elo ratings of Murray and his opponents at the time of their match.

The chart below shows the difference in Murray’s match outcomes compared to expectation, expressed as a percentage (-100 being the worst possible and +100 being the best possible). This is like advanced stats that look at performance over what was expected, such as Points Over Expected used in some team sports.

The wins from expected for Murray in 2017 shows that each loss suggested under performance of shocking levels. That is the most obvious contrast with 2016, where loses to Nadal and Djokovic in his warm-up to the French Open weren’t inconsistent with expectations. Murray’s wins are also revealing. He hasn’t had any win in 2017 on clay (so far) that equals his best wins in 2016.

Looking at the total score of the wins over expected makes it clear that it has been a shocking run for Murray on clay in 2017. And unlucky draws aren’t to blame.

In an interview after Murray’s loss to 20 year-old Borna Coric in Madrid, he was asked for his feelings about his game. ‘I definitely think I need to be concerned,’ he told reporters. Looking at the numbers around his win-loss clay performance in 2017 compared to 2016 backs that feeling up. But it doesn’t tell us *why* Murray is experiencing a decline.

Given the importance of the serve to winning in men’s tennis, we can ask whether a drop on serve is the cause. Murray has already missed some of the 2017 season due to an elbow injury and any nagging effects from that might result in a drop in his service stats.

We can look to his service percentage won on clay in 2016 and 2017 through the Madrid Masters. However, the raw percentages are difficult to interpret because they will be influenced by the return ability of Murray’s opponents, who won’t be exactly the same in each year. We need to somehow factor out the receiver ability so that we can isolate Murray’s service performance apart from his opponent’s return performance.

To do this, I have come up with an opponent-adjusted serve percentage. The way it works is that, based on the past two years of ATP matches, I give each receiver a score that is equal to the percentage of points they takeaway from their opponent. Novak Djokovic, for example, causes his opponent’s to drop 10% on serve, on average. We can call this the “receiver effect”.

I then add the receiver effect to the actual serve percentage to get the adjusted score. So a player who won 55% of service points against Djokovic would have an adjusted serve performance of 65%, indicating how well he actually performed given Djokovic’s strong return ability. For players who actually help a server by playing poorly on return, the receiver effect would subtract from the actual serve percentage won.

The chart above shows Murray’s adjusted serve performance on clay in 2016 and 2017 up to the current time of the calendar. While he was typically performing at a level of 68% on serve in 2016, that has dropped to 63% this year. In 2016 he had 3 matches in which he performed over 75% on serve and only 2 where he was serving below 60%. In 2017, Murray has had only one match performance on clay where he performed over 75% on serve (Madrid vs Marius Copil) but had 3 matches where he has served under 60%.

These numbers show that, this year, Murray is clearly not performing to his serve ability on clay in 2016. This analysis begs the question of what, if not a slow recovery from his elbow injury, could be the cause behind Murray’s drop in serve ability this season?