October 25, 2005
Call it a Comeback
Best Rebound Seasons
With the recent publicity around the comeback players of the year, Ken Griffey Jr. and Jason Giambi, I thought it would be interesting to look at an objective way to pick these awards. So, I set up a basic structure for finding players who had established a notably high level of performance, then suffered a setback in performance and rebounded to be solid contributors again.
This bounceback comes in three steps: the first peak, the valley and the second peak. In order for a player to qualify for our 2005 Objective Rebound Award (or ORA, because we love acronyms and we're hoping that the winner has that special something about him), the second peak should come in 2005. For the initial run, we're only going to consider players whose first peak came in 2003 and valley came in 2004. Later, we'll open it up to look at larger windows, up to five years from peak to peak. Although the subjective Comeback Awards are given out by league, we'll make no such distinction here, to avoid having to split playing time across leagues.
Overall, the level of the rebound is measured by the distance dropped plus the distance gained back, or (Peak 1 VORP - Valley VORP) + (Peak 2 VORP - Valley VORP). Although this method would leave us open to having some rebounds that appeared large because of one large peak on either end, there are so many seasons in question that the highest rebounds end up having large peaks on each end. Once we start to limit the sample sizes down to three consecutive years ending in 2005, you get some interesting "rebounds." Although we could place limits on these, it would take arbitrary cut-offs, and since it's an inexact science and simply a toy at this point, we can eliminate these by sight as they come up.
Since the nature of pitching entails more fluctuation than hitting, the two are separated in these charts (also perhaps because cross comparison isn't necessarily helpful). In the data, I've included the year of each first peak, the valley and the second peak, in addition to the player's VORP total for that year. I also thought it might be interesting to see the player's age and playing time in the valley season (indicated by a *), to help indicate possible causes for the drop off, so I've included total IP or PA for pitchers and hitters, respectively.
Here is the data for three-year rebounds that covered 2003-2005:
Hitters NAME AGE* YEAR1 PEAK1 YEAR2 VALLEY PLAYING* YEAR3 PEAK2 REBOUND Jason Giambi 33 2003 63.487 2004 4.56 322 2005 58.338 112.705 Richie Sexson 29 2003 58.66 2004 7.297 104 2005 56.158 100.225 Morgan Ensberg 28 2003 43.207 2004 11.416 456 2005 61.516 81.89 Carlos Delgado 32 2003 83.279 2004 41.37 551 2005 71.63 72.169 Alex Rodriguez 28 2003 96.319 2004 62.323 698 2005 99.708 71.381 Derrek Lee 28 2003 51.61 2004 43.325 688 2005 105.896 70.856
The graph makes the trend is easily visible. All six of these guys have huge "V" shapes in their graphs for the period from 2003 to 2005, with the exception of Derrek Lee, who gets on here because his 2003 VORP is higher than his 2004 VORP, and his 2005 VORP is the best in the league and 22nd best since 1972 (this graph also shows six seasons better than his, Alex Rodriguez in 1996, 2000 and 2001, Giambi in 2000 and 2001 and Carlos Delgado in 2000). There aren't too many other interesting stories, as basically all of these hitters were good to great in 2003, suffered setbacks in 2004, and rebounded to different extents in 2005.
Pitchers NAME AGE* YEAR1 PEAK1 YEAR2 VALLEY PLAYING YEAR3 PEAK2 REBOUND Esteban Loaiza 32 2003 77.982 2004 2.763 183 2005 42.156 114.613 Hideo Nomo 35 2003 56.23 2004 -23.199 84 2005 -16.831 85.797 Shawn Chacon 26 2003 13.737 2004 -11.875 63.3 2005 37.382 74.87 Derek Lowe 31 2003 24.436 2004 -11.536 182.6 2005 25.27 72.779 Andy Pettitte 32 2003 31.816 2004 16.161 83 2005 72.478 71.972 Kevin Millwood 29 2003 37.552 2004 9.309 141 2005 52.296 71.229
Here we get some examples of pitchers having huge peaks on one side or another, either Hideo Nomo's free fall from 2003 to 2004, with enough of an increase in 2005 (-23 to -16. SWEET!) to make the list, or Andy Pettitte having a career year in his age-33 season, helping the Astros to the World Series. The true rebound, though, is Esteban Loaiza, who had to have the largest four consecutive years of change in VORP in history, hasn't he? Look at his graph, represented by the black line--what most of us figured as a fluke in 2003 was met with his worst year as a pro in 2004, and then this year he had his second-best season. Wait for it, and next year he's bound to bomb even worse, only to bounce back again. He's an odd bird, suffering from the mysterious "odd-numbered years" syndrome.
In a close race, it seems like Jason Giambi is a reasonable choice, and in this case would likely win both the 2005 ORA and its subjective counterpart. Loaiza is the winner based strictly on the data, and his rebound is certainly notable, so he could be our NL winner if we had such a thing. The problem with distinguishing between leagues is self-evident in Loaiza's case especially, since his 2003 and 2004 years were in the American League and his 2005 was in the National League. If the award is based on rebounds, wouldn't the whole rebound have to be in the same league?
Instead of requiring the 2005 ORA winner to have his rebound in the three-year period from 2003 to 2005, we could allow for more leeway, opening it up to a five-year span, while still requiring that the second peak come in 2005. This gives us a few more data points, but Giambi is still the leader, this time by an even greater margin:
Hitters NAME AGE YEAR1 PEAK1 YEAR2 VALLEY PLAYING YEAR3 PEAK2 REBOUND Jason Giambi 33 2001 114.173 2004 4.56 322 2005 58.338 163.391 Pat Burrell 26 2002 61.672 2003 0.123 599 2005 50.035 111.461 Tony Clark 30 2001 34.292 2002 -13.645 298 2005 44.694 106.277 Rich Aurilia 32 2001 87.387 2004 6.054 450 2005 26.123 101.403 Richie Sexson 29 2003 58.66 2004 7.297 104 2005 56.158 100.225 Jermaine Dye 29 2001 29.342 2003 -17.161 253 2005 35.697 99.362
Giambi gets the added benefit of falling from his 2001 peak--which is the tenth best season by a hitter since 1972--to his 2004 valley. Few get to such heights, and even fewer fall to such depths. Pat Burrell managed to be almost valueless (compared to replacement level) over almost 600 plate appearances in 2003, but was a good player in both 2002 and 2005. Tony Clark basically just managed to be good for a few years, then totally worthless in 2002, and useful again in 2005, having a career year in a very interesting role. Rich Aurilia makes the list because of his mammoth 2001 season, meaning that he didn't have a rebound as much as an incredible career year; he's the best data point so far to indicate the flaws of this selection process--he had one great year, and was otherwise a marginally good player. Richie Sexson's three-year rebound from the first chart gets him on this list; it is interesting to note that Sexson was on a pretty steady development path before 2004, when he tore his labrum and only accumulated 100 trips to the plate. He's basically back on track now for a leveling-off and decline phase that we might expect as he turns 31 next year. Jermaine Dye would have an even bigger "rebound" if he got more credit for his 2000 campaign, his best year as a pro. As it is, he had an obvious downturn in 2003, and his return to offensive usefulness in 2005 lands him here.
Pitchers NAME AGE YEAR1 PEAK1 YEAR2 VALLEY PLAYING YEAR3 PEAK2 REBOUND Derek Lowe 31 2002 81.004 2004 -11.536 182.667 2005 25.27 129.347 Esteban Loaiza 32 2003 77.982 2004 2.763 183 2005 42.156 114.613 Chris Carpenter 27 2001 33.167 2002 3.742 73.333 2005 68.519 94.202 Kyle Farnsworth 26 2001 26.938 2002 -19.126 46.667 2005 26.114 91.304 Hideo Nomo 35 2003 56.23 2004 -23.199 84 2005 -16.831 85.797 Roger Clemens 39 2001 54.308 2002 25.639 180 2005 80.626 83.656
All of these guys seem to have one story or another. As pitchers, they go through more variation, and it's easier to see the jumps in the last five years for all of these on this graph. Derek Lowe had an impressive 2002, we've discussed Loaiza, and Chris Carpenter survived injury problems and then absolultely tore apart the league this year. Kyle Farnsworth is an interesting data point because he's the first reliever to appear on the lists, perhaps implying that relievers tend to be more consistent than starters. Of course, since we're measuring based on the peaks, the starters accrue more innings, and therefore have more opportunity to accumulate gaudy VORP totals (as VORP is a cumulative statistic). 2003 was Farnsworth's worst season, as he struggled for the Cubs. They ended up turning to Rod Beck if that helps put it into context. He came back this year and was useful for both the Tigers and Braves, serving as one of the best arms out of a weak Atlanta pen. Hideo Nomo was in the first chart for pitchers, and makes this list because (A) his 2003 was actually good and (B) his fall was really that precipitous. Having Roger Clemens on the list is interesting, but it's mostly due to his recent resurgence after groin troubles hurt him in 1999 and he had another "bad" year in 2002, worth just 25 runs. His return to dominance and ascension to full fledged epic-status over the past couple of years has been well chronicled.
So how do the 2005 fan choices of Giambi and Griffey shake out? Pretty well, actually. We've already seen that Giambi was a fitting choice given his recent track record and his rebound year in 2005. Of course, Griffey had already established stardom before injuries hampered his performance for the first several years of his tenure in Cincinnati. Looking at the numbers, Griffey's best 5 year rebound starts in 2001 and ends in 2005, with his valley in 2002:
NAME AGE YEAR1 PEAK1 YEAR2 VALLEY PLAYING YEAR3 PEAK2 REBOUND Ken Griffey Jr. 32 2001 39.065 2002 9.407 232 2005 59.156 79.406Of course, 2001 was hardly the peak of Griffey's career, as the following graph shows. Basically, the wider the window gets, the further Griffey has fallen. Going back to 1997, arguably his best year at the plate, Griffey went through a steady decline. Luckily, he was able to turn this around starting in 2003, and has improved each season since 2002. This makes him a natural choice for the award--both objectively and subjectively.
It would be interesting to go back through history to find any stretch in which a player goes through consecutive years of decline which ends in a single valley season, and has any number of consecutive years of improvement afterwards. Identifying careers like Griffey's and eliminating Hideo Nomo-style "improvements" would clarify what it means to be a comeback player, and what voters think it means.