March 6, 2002
Doctoring The Numbers
Doctoring the Numbers: Improvement Ratio
In the last installment of DTN, we examined the topic of whether left-handed pitchers take longer to have a breakout season than right-handers do. In the process, we had to define exactly what a "breakout" season is. I used a series of qualifiers to define the term, and it worked pretty well. But there is a much simpler definition:
A breakout season is what Roy Halladay had in 2001.
Sure, a 5-3 record and a 3.16 ERA in 105 innings doesn't look that impressive on the surface...unless you consider that Halladay had a 10.64 ERA in 68 innings in 2000, the worst ERA in major-league history for someone with even 40 innings. Here is a list of the worst single-season ERAs since 1900:
Year Pitcher IP ERA 2000 Roy Halladay 67.2 10.64 1999 Micah Bowie 51 10.24 1973 Steve Blass 88.2 9.85 1998 Andy Larkin 74.2 9.64 1985 Glen Cook 40 9.45
Halladay improved his ERA by 7.48 points in 2001, which--not surprisingly--is a record for anyone with 50 or more innings in two consecutive seasons. What is surprising is the margin by which this is a record:
Pitcher Year 1 ERA Year 2 ERA Diff Roy Halladay 2000 10.64 2001 3.16 7.48 Willis Hudlin 1936 9.00 1937 4.10 4.90 John D'Acquisto 1977 6.54 1978 2.13 4.41 Albie Lopez 1997 6.93 1998 2.60 4.33 Ferdie Schupp 1915 5.10 1916 0.90 4.20
(On a pure percentage basis, it's hard to imagine anyone cutting their ERA by more than Ferdie Schupp, who sliced 82% off his ERA in 1916. He threw just 140 innings in 1916, preventing him from holding the single-season record for ERA.)
It's not just Halladay's massive improvement in his ERA that is so eye-catching, though. After all, looking at Halladay's career numbers, his 2000 season stands out for its overall suckitude. In 1999, Halladay had 3.92 ERA; he gave up just three earned runs in 14 innings in his 1998 cup of coffee, and came within an out of a no-hitter on the season's final day.
No, what makes Halladay's 2001 performance stand out in bold relief from the rest of his career is not his ERA. It's his strikeout-to-walk ratio.
Here are Halladay's walk and strikeout numbers, year by year:
Year BB K 1998 2 13 1999 79 82 2000 42 44 Total 123 139
Moreover, this fits perfectly with the trend Halladay showed in the minors, where he was considered a top prospect despite his maddeningly unimpressive strikeout-to-walk ratios in Triple-A. In 1998, he had 71 strikeouts against 53 walks; in 1997, he had 64 strikeouts against 53 walks. (In fairness, his ratios in the low minors were much better.)
Going into 2000, Halladay's career K/BB ratio in the majors was just 1.13. To put that in perspective, the lowest K/BB ratios of any active pitcher (min: 200 IP) through 2000 were:
Pitcher BB K Ratio Jamey Wright 349 335 0.960 Steve Sparks 307 333 1.085 Jason Grimsley 345 385 1.116 Roy Halladay 123 139 1.130 Dennis Springer 288 254 1.134
Greg Madduxand Bret Saberhagen, these guys are not. These five pitchers include a pair of knuckleballers and a Rocky Mountain refugee, and what a man with Halladay's stuff was doing on this list isn't entirely clear. To his credit, last season he orchestrated one of the most impressive leaps in K/BB ratio in history.
Halladay's ratios, through 2000 and in 2001:
Years BB K Ratio 1997-2000 123 139 1.13 2001 25 96 3.84
That's a remarkable difference. Halladay more than tripled his career strikeout-to-walk ratio in 2001. That's more than remarkable: it's historic. The following chart lists the greatest improvements in K/BB ratio over a pitcher's previous career norms. To qualify, a pitcher had to throw at least 100 innings in his year of improvement, and have at least 200 previous career innings:
--- Season --- ---- Career ---- Pitcher Year Age BB K K/BB BB K K/BB Ratio Hal Brown 1963 38 8 68 8.500 355 589 1.659 5.123 Jack Chesbro 1901 27 52 129 2.481 138 84 0.609 4.076 Cy Young 1904 37 29 200 6.897 931 1734 1.863 3.703 Bill Henry 1959 31 26 115 4.423 156 198 1.269 3.485 Rudy May 1982 37 14 85 6.071 932 1659 1.780 3.411 Red Donahue 1903 30 34 96 2.824 528 438 0.830 3.404 Roy Halladay 2001 24 25 96 3.840 123 139 1.130 3.398
Halladay ranks eighth behind a group of veteran pitchers who suddenly stopped walking people entirely (Cy Young actually qualifies for the list again in 1905, but he was excluded for the sake of brevity.)
There's another way to construct this study. K/BB ratio is a wonderful metric, but it is subject to distortion at the extremes. In particular, a pitcher who is able to cut his walks down to a bare minimum can dramatically increase his K/BB ratio even if his strikeouts don't budge. Consider a pitcher with 20 walks and 70 strikeouts, a ratio of 3.50. If he adds 10 strikeouts to his total, his ratio inches up to 4.00. On the other hand, if he cuts 10 walks, his ratio doubles to 7.00. Hal Brown didn't suddenly turn the corner at age 38. In his "breakout season," he struck out 68 batters in 141 innings, less than one every other inning. But because his control was freakishly impeccable for one fluky season (just 0.57 walks per nine innings), he reached the top of the previous chart.
That's not what Halladay did. Halladay's performance--cutting his walk rate and increasing his strikeout rate--was far more impressive. So instead of comparing strikeouts to walks, let's compare each of them to innings pitched.
Halladay cut his previous career walk rate by more than two-and-a-half walks per nine innings, and increased his strikeout rate by an even higher margin. If we take the weighted average of the two (using an identical process to the one that turns homers and stolen bases into Bill James' Power-Speed number), we arrive at an "Improvement Ratio" of 2.72. This is the highest recorded by any pitcher, at any age, in major-league history.
Pitcher Year BB Imp K Imp Imp Ratio Roy Halladay 2001 2.656 2.787 2.720 Randy Johnson 1995 2.257 2.733 2.473 Todd Van Poppel 1995 3.063 1.976 2.402 Jim Maloney 1963 2.164 2.621 2.371 Charlie Puleo 1987 2.274 2.392 2.295 Duane Ward 1991 1.549 2.999 2.043 Randy Johnson 1993 2.221 1.856 2.022 Jack Kramer 1944 2.622 1.627 2.008 Pedro Martinez 1999 1.368 3.616 1.985 Hal Newhouser 1946 1.458 2.679 1.888
Todd Van Poppelstanding side-by-side with Randy Johnson, Charlie Puleo giving back cuts to Duane Ward...that's, um, a rather eclectic group of pitchers.
In general, the fluke seasons are vastly outnumbered by the truly great pitchers coming into their own. If we eliminate Van Poppel, Puleo, and Jack Kramer (a journeyman pitcher benefiting from the decreased talent level during World War II), we're left with a Hall of Famer (Hal Newhouser), two future ones (Johnson and Pedro Martinez), one of the better starters of the 1960s (Jim Maloney), and Duane Ward, who in the early 1990s might have been the best set-up man in baseball history.
If we extend the list, that trend becomes even more pronounced. Forty-one men have posted an Improvement Ratio of 1.5 or higher. Those 41 pitchers include Kevin Brown, Jack Chesbro, Rich Gossage, Sandy Koufax, Greg Maddux, Sudden Sam McDowell, J.R. Richard, Nolan Ryan, John Smoltz, Mario Soto, Dazzy Vance--and five different iterations of Randy Johnson.
Whatever Halladay started doing differently last year, let's hope he keeps doing it, because he's on to something big. The path he's following has been covered by some of the great pitchers in baseball history. But let's not get too excited yet. Remember, Todd Van Poppel once walked this way, too.