Clutch factor is the percentage difference between a pitchers LevERA+ and his ERA+. Below is a list of the top 15 clutch factors among pitchers with 2000 innings pitched since 1952. There are some very familiar names on the list and some not so familiar names. The most familiar names are of course the two Hall of Famers on the list, Jim Palmer and Tom Seaver. The list also includes three southpaws who are generally regarded as among the greatest lefties of all time but not quite worthy of the HOF: Billy Pierce, Ron Guidry and Vida Blue. There are three active players on the list: Suppan, Colon and Halladay. Among the 202 pitchers who've pitched 2000 or more innings since 1952, these are the only 15 with a positive clutch adjustment factor higher than 4.0%.

1. Steve Trachsel 7.5%

2. Pat Hentgen 5.7%

3. Jeff Suppan 5.5%

4. Johnny Podres 5.3%

5. Jim Palmer 5.3%

6. Darryl Kile 5.2%

7. Pedro Astacio 5.2%

8. Billy Pierce 5.1%*

9. Juan Pizzaro 4.9%

10. Ron Guidry 4.5%

11. Vida Blue 4.5%

12. Tom Seaver 4.4%

13. Al Leiter 4.4%

14. Bartolo Colon 4.2%

15. Roy Halladay 4.2%

Each of these pitchers have an ERA+ that can fairly be said to be deceptively low. Another way to put it is that they were better pitchers than their ERA+ figures might indicate. The fact that Steve Trachsel pitched much more effectively in higher leverage situations may not be particularly interesting, but the significant clutch factors for the more notable pitchers on the above list should be of interest. Five have LevERA+'s well over 120, putting them among the top ranks in this statistic among pitchers with 2000 IP since 1952:

Halladay, 140.1, 3rd behind Martinez and Clemens.

Seaver, 133.7, which ranks 7th

Palmer is 8th with a 132.7

Pierce is 12th at 128.2.*

Guidry is 17th at 124.3.

Some might be moved to ask "why all this attention to LevERA+, Gator Guy?" The answer is simple: LevERA+ is a more accurate measure of pitching prowess than ERA+ because it is more highly correlated with pitcher W-L records. Regression analyses of the 202 pitchers who have pitched more than 2000 innings since '52 reveal that the correlation coefficient between LevERA+ and winning percentage is .726, as compared to .682 for ERA+, a significant difference given that (i) the difference between ERA+ and LevERA+ is generally very small, rarely exceeding 4%, and (ii) nearly half of a pitcher's winning percentage is determined by how many runs he receives in support.

Regression analyses of 617 pitchers who have pitched at least 500 innings since 1952 similarly revealed that winning percentage was more correlated with LevERA+ than ERA+. The larger sample size also permitted analyses that controlled for run support by selecting a data subset that excluded pitchers with extremely good or poor run support. This was done by removing from the data set pitchers with very high or low ratios of winning percentage to LevERA+. After multiplying winning percentage by 1000, the average ratio of winning percentage to LevERA+ was approximately 5.0. There were 437 pitchers with a ratio between 4.5 and 5.5, and the coefficient of correlation between LevERA+ and winning percentage for this subset was .8848. The correlation between ERA+ and winning percentage was .8473.

The correlation coefficients for the 240 pitchers with a ratio of winning percentage to LevERA+ of between 4.75 and 5.25 were extremely high: .9662 for LevERA+ and winning percentage and .9152 for ERA+ and winning percentage. At levels of correlation this high, the difference between the LevERA+ coefficient and the ERA+ coefficient is very significant. For the 240 pitchers with a ratio between 4.75 and 5.25, the R-squared coefficient of determination for LevERA+ was .9336, meaning that run support and other factors accounted for less than 7% of the variation in winning percentage. The corresponding R-squared coefficient of determination for ERA+ was .8376.

The findings with respect to the correlation between LevERA+ and winning percentage are consistent with regression analyses of the primary set of 617 pitchers that showed a correlation between clutch adjustment factor and winning percentage of .2697 and an R-squared of .0727.

The regression analyses described above establish conclusively that LevERA+, calculated on the basis of the win probability-based "clutch" statistic at B-R.com, is a more accurate predictor of pitcher winning percentage than ERA+.

______________________________