Insight: Causal inference can help uncover which tactics are effective. But it’s important not to test too many hypotheses.
I’ve always admired players who overcome adversity. I featured one such player, Andy Murray, in the previous series on the 2016 Wimbledon Men’s Final between him and Milos Raonic. Two additional such players are Serena Williams and Caroline Wozniacki. Williams made four Slam finals since returning to the tour in 2018. But even though she’s lost all four of them and remains one Slam away from tying Margaret Court’s all-time record, she’s still going for that record.
Meanwhile, Wozniacki’s career took off when she finished year-end #1 in 2010 and 2011. However, she did not win any Slams, and as time went on many people thought she would never win one. But exactly six years after she lost the #1 ranking, she regained it in January 2018 by winning the Australian Open, her first Slam. Thus, she set the record for regaining the #1 ranking after the longest time without it. (Update: Wozniacki is retiring from tennis after the 2020 Australian Open. Congratulations on a wonderful career!)
Furthermore, just as the Murray-Raonic match was the last time Murray won his hometown Slam, the 2014 US Open Women’s Final was the last time Williams won hers. It was also the most recent meeting between her and Wozniacki in in a Slam. In just 1 hour and 15 minutes, Williams defeated Wozniacki 6-3, 6-3 thanks to big serves (7 aces in 9 service games), aggressive returns (4 return winners), and control of most rallies (29 winners to Wozniacki’s 4).
In the previous series, I focused on the importance of causal inference, or figuring out why statistics are the way they are. I uncovered an offensive tactic with a deceptively high win probability, a defensive tactic with a deceptively low win probability, and a tactic whose win probability was deceptive for one player but not the other. In this series, I use the Williams-Wozniacki match to show how causal inference can help players strategize to win matches.
In light of how well Williams played, I analyze this match from Wozniacki’s perspective. Can the match data reveal ways in which Wozniacki could have won more points? To answer this question, I first identify tactics that gave her a relatively high win probability, such as down the middle returns. While doing so, I guard against testing too many hypotheses, lest I uncover false-positive results. Then I contextualize those tactics and quantify whether they were worth employing more often or not.
