Insight: Despite a lower win probability when playing from behind the baseline, Wozniacki should not have changed her court position. But she should have hit fewer backhands down the line.
In this article, I use causal inference to identify ways in which Caroline Wozniacki could have won more rallies against Serena Williams in the 2014 US Open Final. However, searching the data for answers, also known as data mining, is dangerous. If you search long enough, some specification will appear significant, even if the data contain no meaningful information. Furthermore, while causal inference can expose spurious correlations, it cannot protect against data mining. For this reason, qualitative knowledge of the players comes in handy because it allows me to limit my search to a smaller number of specifications. In turn, the results are more likely to carry over beyond the current match, which means they can help players strategize in future matches.
Before charting a match, I like to watch it as a normal observer to get a sense of the big picture. One thing that stood out to me in this match was Wozniacki playing defense from up to several feet behind the baseline. Essentially, she was ceding control of rallies and hoping to win by attrition. Was that a good strategy, or should she have stepped into the court to control more rallies herself? Causal inference is necessary to answer this question because, as I showed in Part 2 of the Murray-Raonic series, defensive tactics have inherently lower win probabilities than offensive or neutral tactics.
To test whether playing from behind the baseline helped or hurt Wozniacki, I regress WIN on BEHIND, which equals 1 if Wozniacki’s entire body was behind the baseline at contact, for all rally shots. I do so with and without two of the three controls defined in the Murray-Raonic series: SHORTt-1 and CENTEREDt-1. However, I omit DEFENSIVEt-1 because, as shown in Table 1, she only received two defensive balls all match. Per Table 2, BEHIND’s coefficient is negative without controls, positive with them, never greater than 10%, and never significant. Therefore, she was no less successful playing from behind the baseline than from on or inside it. As such, stepping up in the court likely would not have improved her performance.
However, a different tactical adjustment may have helped Wozniacki play better. Earlier in the tournament, she broke down Maria Sharapova’s forehand with down the line backhands, per an article by Craig O’Shannessy. She then continued this strategy against Williams, hitting 37 of 76 rally backhands down the line instead of crosscourt, per Table 1. I investigate whether this strategy was effective in the final by defining DTL = 1 if her backhand landed anywhere in Williams’s forehand half of the court. I also define an additional control variable OPEN = 1 if Williams’s previous shot was hit from the backhand half, leaving the forehand half open. Unfortunately for Wozniacki, her down the line backhands did not work against Williams as they did against Sharapova. As shown in Table 2, they were over 20% less effective than her crosscourt backhands. This result is significant at the 10% level both with and without controls.
By qualitatively determining ahead of time to analyze Wozniacki’s court position and backhand placement, I was able to learn two pieces of information: playing from behind the baseline was not an ineffective tactic but hitting down the line backhands was ineffective. Unfortunately, this result does not mean that hitting fewer backhands down the line would have been enough to reverse the outcome. Having said that, in Part 2, I will identify helpful serving and returning strategies for Wozniacki, along with closing thoughts on how players can improve their chances against dominant opponents in future matches.
Table 1: Summary Statistics
Table 2: Win Probability Regressions
