Pitching Stats
Who is Scoresheetwiz?
Archived Commentary
Most Useful Sites
Archived Analysis
Quick Links (revised)

finding the best pitching stats (part I)

Finding a pair of stats for pitchers that has the simplicity and accuracy of adding OBA (OBP) and SlgA for hitters might be the Holy Grail of stats these days. Most fans use some combination of ERA, WHIP, K:W, and HR/9. Of course, that’s excluding the old fashioned less sabermetricly inclined fans still looking at W-L records and Saves. The most serious studies on this issue apparently agree that K:W and HR/9 do the job the best. However, I have a serious issue with HR/9. Home runs come too infrequently. It is an inaccurate stat due to this small “N”. Further, ball park factors are probably more volatile with Home Runs than with other baseball occurrences.


After the 2003 season, I started using Slugging Average Against. ESPN provides the stat. I guessed that counting home runs four times a single, triples three times a single, etc. averages out the lucky hits, but contains a large enough sample to be significant. Ken Warren author of Ballpark Figures (the most accurate player stat projections on the market) suggested that I should, at least, take Batting Average away from Slugging Percentage. This stat called was coined Isolated Power by Bill James, I believe. Effectively, it takes away all singles and counts home runs three times the value of a double.


After a quick review of veteran pitchers starting with “L” (Leiter, Leiber, Lidle, Lilly, Lima, and Lowe - the discussion sprung from a discussion about Derek Lowe), I discovered the Isolated Power Against or SLgAA – BAA did appear to be the most effective stat in combination with K:W in matching a player’s ERA, at least, more so than SlgAA alone.


So, I began to expand the study. On an Excel spreadsheet, I rounded up stats for all the veteran pitchers from A-F and listing their lifetime K:W, K/9, HR/9, SlgA.A., SlgA.A.-BA.A, and ERA, plus the median deviations of each statistic. While, I was at it, I took down a pitcher’s K:W, K/9, and HR/9 for his AA seasons, his AAA seasons, and his career minor league seasons. If the pitcher had less than 75 at AA or AAA, I entered the numbers in blue instead of black. If he had less than 35 innings to look at, though, I left it blank.


For the median deviation, I looked at all their seasons of pitching 75 innings or more. If they pitched an odd number of seasons, I listed how far away the stat was from their career stat for the particular season which was the middle season in difference. In other words, if a pitcher’s K:W for his 5 seasons of over 75 innings are 2.20, 2.30, 2.40, 2.50, and 2.60. And his career K:W is 2.41, then his deviations are .21, .11, .01, 09, and .19. The two closest seasons are ones with .01 and .09 differences. The two furthest away are the seasons of .21 and .19 deviations. Hence, the median deviation is .11. If the pitcher had an even number of 75+ seasons, then I averaged the two middle seasons of deviations. The point of finding the median deviation was to get a sense of how reliable the stat is. If it is something which swings wildly from year to year, it probably isn’t as reliable as a stat that is more stable. There are probably better, but more complicated ways of measuring this. If anyone wants my Excel sheet for that or their own study for any reason, I’d be happy to send it to them. Just drop me a note:


For each median deviation, I additionally calculated the percentage of the stat those deviations represented. Instead of totals for each stat, because each of the stats transcribed are qualitative, not quantitative, I show the average of each stat.


I haven’t had time to do an in depth analysis, yet. Perhaps, more pitchers are needed to get good numbers. I do include several recently retired pitchers. Eligibility rested with whoever has their Slugging Against stats in the ESPN data bank. Undoubtedly, the data is skewed towards what is true for pitchers who had at least of few seasons in their team’s rotation.


At first glance, though, my hypothesis holds up. Isolated Power is a much more stable statistic than Home Runs per nine innings. And it looks at least as accurate, if used as a combo with K:W. However, SlgA-BA is the one stat where I took a shortcut by calculating the median deviation as the Slugging Against median deviation multiplied by the percentage that each pitcher’s opponent’s SlgA-BA is of their SlgA.A. If that is inaccurate, I doubt it  is off by much, while HR/9 had comparably huge deviations.


When I do further investigations in the data, I’ll update this page. In the meantime:


Bartolo Colon was surprisingly the most consistent of the pitchers studied. Name the stat. He’s had a couple of way-off years, but most of his years have been close to identical. Four of his seven seasons have been within .30 of his career ERA. After his rookie season, his K:W have been 2.00, 2.12, 2.16, 2.23, 2.16, 2.58, 2.23, 2.31.  His Slugging Against Averages: .447 (rookie year), .379, .398, .371, .412, .386. .402, and .472 (last year).


Jeff Fassero has the most divergent career in the study. That is because for half of his career, he was a terrific pitcher. Since he “lost it” in 1999, he has managed to hang on and have quite a few 75+ inning seasons of minimal major league quality.


Getting into the minor league stats, Chuck Finley stands out as an argument against the adage that pitchers need to be brought along cautiously. Although, he is a college guy, he started his post college year in the rookie Northwest League. The following year, he pitched a mere 12 innings in the lower A level Midwest League before he was promoted all the way to the Angels, where he finished the season pitching effectively in relief. After one season as a long reliever / spot starter, he became a fixture in the Angels rotation. By his second year in the rotation, he was the Angels’ ace. Smartly, though, the Angels kept him under 200 innings until he was 27. Suddenly pitching 236 innings may have been too large of a jump, however, as that was his best season. (Over-increase in workloads is another of my pet theories about pitchers. I believe it is more important that watching pitch counts or total innings in a season. However, that’s for another study.)

John Carter