Mechanical Adjustments Are Usually B.S. by Dave Cameron June 23, 2008 After his May 21st drubbing against the Detroit Tigers when he allowed nine runs in just over two innings of work, Jarrod Washburn’s ERA stood at 6.99. He wasn’t actually pitching much worse than usual, however, posting a 4.75 FIP that was basically in line with his career averages. He was just struggling to leave runners on base, and the runs were being piled up due to a poor performance with men on base. With an ERA over 2.00 points higher than his FIP, Washburn was a prime candidate for regression to the mean. Not surprisingly, in his last five starts, Washburn’s ERA has been reduced significantly. His ERA during his last five starts is 3.10 after a solid performance against the Braves yesterday, and Washburn is crediting his college pitching coach. “It was a mechanical adjustment I made after I called my college coach and asked him what I was doing wrong,” Washburn said. “He knew. It was a little adjustment at the beginning (of the windup) that turned into something big by the time I released the ball.” Pitchers do this all the time – they struggle, they make some minor change, and the struggles end, so therefore, the minor change fixed the problem. Unfortunately, it’s almost never true. Here’s Washburn’s performance up to the phone call and since, broken down by metrics that actually judge pitching effectiveness, rather than a useless measure such as ERA. April 4th – May 21st: 47 2/3 IP, 1.89 BB/9, 5.67 K/9, 11% HR/FB, 4.75 FIP, 4.93 xFIP May 25th – June 21st: 29 IP, 4.97 BB/9, 6.21 K/9, 4% HR/FB, 3.92 FIP, 4.98 xFIP If we were going to evaluate the usefulness of Washburn’s mechanical tweak based on a ridiculously small sample, we’d be forced to conclude that it destroyed his ability to throw strikes and was otherwise pretty useless. The uptick in strikeout rate is basically meaningless (it’s two strikeouts total over five starts), and the decrease in allowing runs is due to one very obvious unsustainable performance – the home run per fly rate. Eight of Washburn’s 72 fly balls went over the fence in his first ten starts, but just one of 28 has gone for a home run in his last five starts. That’s not the result of a mechanical change – that’s random variation that has nothing to do with Washburn. Jarrod Washburn isn’t pitching any better since he called his college pitching coach – you could easily make an argument that he’s pitching worse. However, because his ERA has predictably regressed to the mean, we get fed stories about his supposed improvements and the cause of those improvements, all of which are bunk. Tipping pitches, holding the glove higher, changing grips – it’s almost always a post-hoc explanation for regression to the mean, and 99% of the time, it’s got no grounding in reality. Jarrod Washburn’s the exact same guy he was a month ago, and this entire non-story is simply another reason why results based analysis is doomed to failure.