True talent: Wells, Sheff, Hurtby Sal Baxamusa
February 25, 2008
Projections are often thought of as a tool for general managers, fantasy or real, that predicts what a player will do in the upcoming season. While preseason prognostication is a good reason to use player projections, projections are, at their essence, an estimate of a player's true talent at any given time. And we can make that estimation at any time—in April, in August, in January—so long as we incorporate all of the latest performance data. The myriad preseason projection systems are really telling us what they consider to be players' true talent at this moment in time. If we want to know Albert Pujols' true talent as a hitter on any given day, even in the middle of the season, we can use all of his available performance data to create a projection. As a hitter accumulates plate appearances, his performance data change and so too does his projection, allowing us to take a dynamic look at a player's true talent.
Last time, we looked at using projections to understand how a player's true talent changes over time, and in particular, within a season. As before, we'll be using the Marcel forecasting system to create these projections. You can read more about Marcel here.
With that in mind, let's check out a few hitters and see how—or if—their talent has changed over the last few years.
Vernon Wells and Andruw Jones are (superficially) similar players: bevy of gold gloves, average OBP, good pop. Wells, however, had the good fortune to sign a contract extension after his banner year in 2006. Good timing, too, since he was awful in 2007. Which is the real Vernon Wells?
The plot on the right begins on April 1, 2005 and runs through Oct. 1, 2007. Each letter on the horizontal axis represents a month in the baseball calendar: April, May, June, July, August, September, October, and then April again. The blue line at any given date is Wells's OPS over the prior 365 days. The red line is his Marcel projection, or true talent, on that day.
On April 1, 2005, Wells had a Marcel-projected OPS just south of .850, and his projected OPS hovered in the .830-.850 range over the course of the season. Despite average-ish numbers, his projection remained strong, mostly on the strength of a strong 2003 and the fading memory of a weak 2002, plus an upward bump for being young. He ended 2005 with a .690 OPS in August and September, which pushed his projection down to around .825. Wells was on fire most of 2006 as he posted a .900 OPS. But you'll notice that his talent level (projected OPS) didn't change very much. Why?
Part of the reason is that his middling 2005 performance was dragging down his projection, despite posting very good 2006 numbers. And apparently Marcel thinks that there is nothing unusual about a true talent .825-.850 hitter posting an OPS of .780 in one season and .900 in the next.
Wells' 2007, however, was a different story. The Blue Jays center fielder struggled mightily, including a brutal post-break .639 OPS. (Was Wells injured? Marcel doesn't know anything about injuries.) You can see the blue line taking a nosedive, indicating how poor his performance was. And since recent performance is weighted most heavily by Marcel, his projected OPS also took a nosedive, enough to push his true talent OPS south of .800. Still, considering that he barely cracked a .700 OPS in 2007, it may be somewhat surprising that our estimate of his talent isn't any lower.
Gary Sheffield, Hall of Famer? Could be. He's closing in on 500 home runs, has fewer career strikeouts than walks, has posted a career OPS+ of 143, and has played in over 2300 games. If he had been a little nicer to the media during his career, they might already be touting him as a no-doubter. As it stands, he might need a few more productive seasons. But his career trajectory over the last few years suggests an uphill battle.
I haven't done an exhaustive search, but I would guess that few players have seen such a precipitous drop in true talent over the last few years. In early 2005, Sheff was still the fearsome bat-waggler that we all know and love. Entering 2005, his true talent as a hitter was about a .910 OPS. Sheff delivered a very good year at the plate, producing a .890 OPS. His true talent OPS took a modest 30-point hit, and in April 2006 his true talent OPS was .880, still very good.
In 2006-2007, Sheff saw a steep decline in performance, and he missed a lot of time; now, injury, age and performance have decimated our estimation of his true talent. He's now projected as a true talent .800 OPS hitter, over 100 points lower than his talent entering 2005. I would venture to guess that such a precipitous drop is common for late-career players, although I'm not sure.
Of course, we're not limited to only the last few years. We can run a Marcel projection for any player, at any time in his career. For example, check out the career of one of the best right-handed hitters of his era: Frank Thomas. As before, the blue line shows his actual performance over the previous year, while the red line shows his true talent. The heavy black lines separate the seasons, and the labels indicate the beginning of a season.
The neat thing about the "true talent" trace on this plot is that we can consider it to be his true career trajectory. That is, we are viewing how Thomas' talent changed over the course of his career. So, for example, the Big Hurt's hitting chops didn't peak during his incredible 1994 campaign, which is what you might guess simply by looking at the performance data. When his performance is translated into talent via proper weighting and regression, we can see that his peak occurred on June 9, 1996. On that day, his true talent OPS was an incredible 1.071.
And, since players aren't just OPS-generating automatons, we can view their career trajectory in a number of different lights. Below you can see the rise and fall of Thomas' raw power, as measured by ISO (isolated slugging, which is the difference between this slugging percentage and batting average). His true talent power rose steadily early in his career and peaked in the mid-90s. Big Frank's doesn't have the raw power he had at his peak, but his modest career renaissance over the last few years has coincided with a modest bump in his true talent power.
Younger fans may not recall, but Frank Thomas used to hit for very high averages. His career arc for batting average is shown below. From 1992 to 1997, he was a pretty decent bet to contend for the batting crown. You can't tell simply from his up-and-down post-1998 performance, but his ability to hit for average saw a slow decline through 2002, "collapsed" (relatively speaking) in 2003-2004, and has held steadily in the .250-.260 range since then.
If these sorts of career arcs interest you, THT alum Dan Fox recently did similar work on Hank Aaron and (gulp) Barry Bonds.
Next time, we'll catch up on a few other hitters. If you have any requests for players that you'd like to see presented this way, let me know.
Sal Baxamusa is a graduate student in chemical engineering. He can be reached here.
<< Return to Article