Dave Righetti: Lord Of The HR/FB Rate
In a couple of recent posts, I created and then tested a regression model which helps explain the variance in home run per fly ball rate. I ended up with a model which performed really well, so it’s time to turn it loose on the question that sent me down this path in the first place: can Dave Righetti really coach his pitchers to a better HR/FB rate? It turns out, the answer may be “yes”, and it could be more emphatic than I would ever have guessed.
Up until this point, the model included data from 2002-2009, leaving last season out of the sample for testing purposes. Now that I’m satisfied with the validity of the model, I’m going to throw the 2010 data into the sample to increase the sample size and get the most accurate coefficients as possible.
The first instinct for testing this hypothesis might be to compare the model’s projections for Righetti’s Giants against the league average. But since there have been countless manager and pitching coach changes in the league since 2002, comparing Righetti against the aggregate performance of the rest of baseball might only prove that having a consistent coach produces better results. After all, a pitching coach who has held his job for nine-plus years must be doing something right. Instead, Righetti will be tested against the four other pitching coaches who have been with their teams since 2002. Those coaches are the Cardinals’ Dave Duncan, the White Sox’s Don Cooper, the Cubs’ Larry Rothschild and the Twins’ Rick Anderson. True, Rothschild is now with the Yankees, but since the time frame is 2002-2010, he is safe to include.
The first test is to see how many of each coach’s starters outperformed the HR/FB rate that the model would predict. Using the same data restrictions from when the model was built (80-plus innings, same team all season), each coach has between 40 and 48 qualified pitcher-years in the sample, meaning one pitcher’s performance for one season. The question is simply: in how many of those instances did the pitcher outperform his expected HR/FB rate?
For the most part, the model splits each coach’s qualified pitchers in half between those who outperformed their expected HR/FB rate and those who underperformed their projection. Righetti is the only exception. Of the 44 qualifying pitcher-years in Righetti’s reign with the Giants, 38 had a lower HR/FB rate than the model predicted. Considering none of these other coaches had even 60-percent of their pitchers fall on one side or the other, Righetti’s 86-percent rate of starters outperforming their expected HR/FB is simply stunning.
Let’s take things to the next level. Instead of looking at only qualifying pitchers in select seasons (five of Righetti’s pitcher-years were Matt Cain), let’s look at every inning thrown by a starter over the past nine seasons. By plugging data from the multiple season database into the model, we get an expected overall HR/FB rate for the entire time period. Since park factor is one of the variables in the model, a weighted average of park factor was used for the nine seasons. This is particularly important for Anderson’s Twins since they moved ballparks during the time frame.
Again, Righetti is the only pitching coach who seems to have any significant effect on HR/FB rate, and it’s decidedly negative. His starters have a HR/FB almost two standard deviations away from the model’s expectations, while none of the other coaches are over one standard deviation in either direction. Also note that this looks at more than 8,700 innings under each coach’s tutelage. To outperform the model’s expected rate, which predicts the Giants to have a below-average HR/FB already, seems beyond the bounds of luck.
For one last test, let’s include relief pitchers into the sample. Until this point, relievers have been excluded due to their low sample of innings and a general propensity to have low HR/FB rates. But when looking at an aggregation of nine seasons of data, sample size is no longer an issue. It stands to reason that if Righetti can coach good HR/FB rates, he would let his relievers in on the secret as well.
Game, set, match, Righetti. For 13,000 innings over nine seasons, his pitchers have outperformed their expected HR/FB rate by 1.5-percentage points, an effect none of the other four celebrated pitching coaches even came close to matching. Granted, HR/FB rate does not in itself equate to pitcher success, and it may not even be a priority for other coaches. For example, Duncan is notorious for his pitchers’ ground ball rates and Anderson for his staff’s low walk rates. But the question at hand was “Can HR/FB rate be coached?” From the looks of it, the answer may be, “Yes, but only if you are Dave Righetti.”
Doing some quick math shows the impact of the Righetti effect. The Giants have faced 55,874 batters from 2002-2010. Subtracting strikeouts, walks, intentional walks, hit batters and errors results in about 39,000 balls in play. Over that time frame, the Giants had a 37.7 FB%, meaning there were about 14,700 fly balls hit. If the model’s 10.1 HR/FB rate for San Francisco was correct, then Giants’ opponents would have hit about 1,500 home runs, but opponents hit only 1,271 home runs. The difference is over 200 home runs, or about 25 per year.
Considering an average home run is worth 1.42 runs and every 10 runs is roughly equivalent to a win, the team’s home run prevention has contributed about 30 wins to the Giants since 2002, or about three per season. If we were to give Righetti all of the credit for that difference based on an assumed ability to coach HR/FB alone, much less any effect from improving his pitchers’ traditional skills such as strikeout, walk, or ground ball rates, than Righetti would have created about $110 million in value for the Giants over the last nine years.
It’s unlikely that the difference is all Righetti. We may be underestimating the park factor, or the Giants may target pitchers who can succeed specifically in their park. There is room for a lot of good luck in there as well. But, given that Righetti is one constant in a sea of ever changing variables, and the results continue to stay the same year in and year out, it’s likely that he is part of the answer. We probably need to start including him in discussions about the best pitching coaches in baseball.
Jesse has been writing for FanGraphs since 2010. He is the director of Consumer Insights at GroupM Next, the innovation unit of GroupM, the world’s largest global media investment management operation. Follow him on Twitter @jesseberger.
One way to test whether or not it’s Righetti or the pitchers they acquire is to look at pitchers who pitched for other coaches before or after pitching for Righetti. It cuts down the sample a touch, but you could realistically include relievers too, because his adjustments probably wouldn’t be applied just starters.