<
>

Statistically, nerves were a factor in Serena's loss

NEW YORK -- After losing her bid for a calendar-year Grand Slam on Friday, Serena Williams insisted she did not succumb to the magnitude of the moment.

"No," Williams said in her postmatch news conference. "I told you guys I don't feel pressure. I never felt pressure. I never felt that pressure to win here.

"I said that from the beginning."

But in the end, she lost her US Open semifinal 2-6, 6-4, 6-4 to Roberta Vinci.

"I made a couple of tight shots," Williams conceded, "to be honest."

The numbers suggest there were far more tight shots than that.

On the day before the semifinal, Benjamin Tom, a Los Angeles analytics expert, broke down Williams' Grand Slam run for an ESPN.com story. He isolated "clutch stats" and demonstrated that Williams' game actually elevated in critical moments, such as tiebreakers, break points, deuce and advantage points.

Tom ran the numbers for the loss to Vinci and, for the first time in 13 matches at Wimbledon and the US Open, Williams' performance declined appreciably in clutch situations in nine of 11 categories:

  • First-serve percentage in non-clutch points was 65 compared to 56 for clutch points.

  • The ratio of winners to unforced errors went from 1.3 to .9.

  • Unforced errors rose from 23 percent to 30 percent.

    Moreover, Williams' third-set numbers were startling because it's human nature to constrict in those moments and play more poorly.

    "She exhibited all of the signs of a player who is feeling the pressure," Tom explained. "Her racket speed seemed to come down, and her movement fell off. Overall, the numbers from this semifinal are quite remarkable."

    Williams' third-set statistics -- which in the 11 previous Grand Slam matches that went the distance improved dramatically -- were way off. Williams, measured against the totals from the entire match, declined in seven of eight categories.

    Of particular note were the fall-offs in second-serve winning percentage (45 to 36), receiving points winning percentage (41 to 33) and, most telling, the ratio of winners to unforced errors (1.25 to .79).