April 16, 2009

Checking the Numbers

Keeping Rare Company

by Eric Seidman

Printer-friendly

Contact Author

While revisiting the 2008 dominance of Cliff Lee last week, we investigated how pitchers with similarly large spikes in ground-ball rates have fared in subsequent seasons. The results weren't pretty, and showed that massive rate increases in this area have been few and far between since 1954, and a very low percentage of these pitchers have been able to sustain these higher rates. The research in no way invalidated Lee's success, but rather suggested that factors outside of a change in approach could have an influence on his 2009 performance. Hurlers intent on inducing grounders tend to follow a different set of rules when it comes to HR/FB rates and their percentage of unearned runs than do their fly-balling colleagues, and the group relies on defense more than those whose skill is missing bats. Combine all of these ingredients, and it becomes evident that even if Lee were to become just the fourth post-1994 pitcher to increase his ground-ball rate by at least eight percent and then see it fall no lower than three percent over the following three seasons-all while meeting respectable playing time qualifiers-he is still not guaranteed even a fraction of the success he experienced last season.

Lee's ground-ball rate had a performance-altering accomplice last season in his unintentional walk rate. Following a 3.24 UBB/9 in 2007, Lee pared his free passes to a Maddux-esque 1.33. Calculating the UBB/9 in my database involved the prerequisite coding of innings pitched, and in order to keep everything consistent for the purposes of this article, I re-ran the ground-ball rate queries with total innings as opposed to the batters-faced numbers used last week. Since Lee threw just over 97 innings in 2007, the minimum was set at 95, requiring a pitcher to have thrown at least that many innings in the year of the rate increase and the year prior in order to be included.

From 1954-2007 there have been 1,123 unique pitchers who threw 95-plus innings in consecutive seasons on at least one occasion. This group actually produced 5,031 such instances, since many of the pitchers satisfied the innings criteria on multiple occasions. Looking for ground-ball rate increases of at least 8 percent from one season to the next among all of these cases returned 254 unique pitchers and 288 total occurrences (some accomplished this more than once). These rate spikes have occurred in just 5.7 percent of the 5,031 total instances during this span of over 50 years.

Moving forward one year, 99 pitchers (in 106 pitcher seasons) of the remaining 254 had rate spikes and were able to maintain them while meeting the innings-pitched benchmark. Finally, 59 (in 61 pitcher seasons) of the 99 pitchers capable of sustaining the rate in the year following the increase were once again able to record ground-ball frequencies in the same general vicinity. The rising percentages here can be a little misleading; few pitchers in the sample were able to establish the change as a new skill. Simply put, a pitcher has been able to boost his ground-ball rate as significantly as Lee had last season and then post equable rates over the following two seasons in just 1.2 percent of the situations in which they pitched over 95 innings in consecutive seasons.

How do unintentional walk rates fare in this regard? Lee cut his UBB/9 by almost two full walks last season, but since similar occurrences are incredibly rare, we'll adjust the minimum drop-off to one and a half free passes. The larger sample of 5,031 instances meeting the innings-pitched qualifier is still intact, and just 131 pitchers were able to cut their unintentional walk rate this much from one year to the next (143 instances). Of those 131, 60 were able to either shave their walks even further in the following season or regress by no more than 0.3 walks per nine innings (65 instances). Extending out one more season, 27 of the 60 who significantly reduced their walk rate and then held steady in the next season were again able to keep their UBB/9 in the same neighborhood as the first year (30 instances), so that less than one percent of the overall sample were able to reduce their walk rates as much as Lee had last season, and then repeat the new skill over the following two years.

Suffice to say, increasing a ground-ball rate or decreasing a walk rate to the extent of Lee's statistical shifts last season is very rare, and even rarer are the number of pitchers who were able to establish these new rates as legitimate skills. Lee, however, accomplished both of these changes in the same season. The data mentioned throughout this article are independent of one another, but what happens if the two rates are combined? Between 1954 and 2008, only 10 pitchers have reduced their walk rate and increased their ground-ball rate to such an extent in the same season, and two of them happened to accomplish the feat last year. In addition to Cliff Lee, there was another pitcher whose effort went largely unnoticed in '08: Carlos Villanueva of the Milwaukee Brewers.

Below is the complete list of the 10 pitchers and their rates. (Keep in mind that these ground-ball rates, courtesy of Retrosheet, are different than those offered by Baseball Info Solutions that can be found on other sites.)

Player             Years   GB1  GB2  UBB1  UBB2
Ray Lamb          1971-72  .49  .61  3.52  2.01
Mike Norris       1979-80  .48  .56  5.23  2.56
John Denny        1982-83  .66  .74  4.48  1.78
Al Leiter         1997-98  .40  .49  5.17  3.22
Dwight Gooden     1999-00  .35  .47  5.01  3.51
Steve Sparks      1999-00  .43  .52  5.00  2.51
Kevin Appier      2000-01  .43  .52  4.24  2.61
Daniel Cabrera    2006-07  .41  .49  6.26  4.49
Cliff Lee         2007-08  .36  .45  3.24  1.33
Carlos Villanueva 2007-08  .37  .48  3.94  2.41

Of the eight pitchers other than Lee and Villanueva, six logged 95 innings or more in the season following their remarkable rate changes. Only Daniel Cabrera and John Denny fell within the +0.30 maximum UBB/9 increase, and only Cabrera and Steve Sparks produced similar ground-ball rates. Cabrera is the only other qualifying pitcher we have on record who had these changes and sustained them in consecutive years, and as we discussed prior to the season, Cabrera now appears to be on the last legs of his career; he may struggle to hold down his rotation spot for the lowly Nationals in the coming months.

Many analysts, myself included, chalked up Cliff Lee's tremendous 2008 season to his change in approach, as he attacked hitters in order to limit free passes and pitch to contact. When his luck-based indicators normalized, it became clear that Lee's season was no fluke. What we failed to take into account was the rarity of such drastic changes in both unintentional walk and ground-ball rates. Lee, and Villanueva for that matter, may become precedent-setters if these rates can be sustained this season and beyond, but history is certainly not on their side. Granted, past performance does not always accurately project future results, but we have all underestimated the rarity of Lee's and Villanueva's accomplishments last season. Even if the new rates stay put, there are several other factors that could drag down their performance lines, but the historical odds are slim to none that they'll be able to avoid regression or fluctuation. To date, Lee's two starts and the three appearances by Villanueva constitute far too small of a sample from which to draw any conclusions, but it will be interesting to track their rates as the season progresses, and monitor their positions relative to last season as well as how they affect the overall results.

Eric Seidman is an author of Baseball Prospectus. 
Click here to see Eric's other articles. You can contact Eric by clicking here