September 28, 2005
Lies, Damned Lies
Beaten Down by PECOTA
by Nate Silver
Will Carroll, Paul Swydan and I were batting around various types of BP-branded All-Star teams on our internal listserv this week--look for more of these coming soon--and naturally the idea of an All-PECOTA team came up. The idea is to identify those players who have most exceeded their PECOTA projections (“Beat PECOTA”), as measured by actual VORP, less VORP as projected in the very last iteration of our preseason depth charts. In some cases, these will be players who had ridiculous fluke seasons, in others, guys who made some important changes in their game, and in others still, players whom PECOTA went short on and simply got burned. Well also want to look, of course, at those players who most underperformed their PECOTAs--what I call the Beaten Down by PECOTA Team.
The PECOTA All-Star teams don’t include any pitchers. Pitchers have unexpected seasons all the time--that’s the nature of the beast--but usually the stories behind them aren’t very interesting: "he was hit-lucky," "he blew out his labrum," "Leo Mazzone." I’d rather explore the hitters in more detail than put out a series of bullet points. I’m aware that the season is not quite over; VORP figures are taken through Sunday night’s games.
Beat PECOTA: Michael Barrett, Cubs (+15.2 VORP differential); Jason LaRue, Reds (+15.2)
Beaten Down: J.D. Closser, Rockies (-21.4 VORP differential)
It’s been a rough year for catchers. One of the more telling statistics that I’ve seen this season is the one that was flashed on the U.S. Cellular scoreboard last Tuesday: "Victor Martinez leads AL Catchers with 74 RBI." Martinez has since increased that number to 78, which is somehow good enough not only to lead the AL, but also all of baseball--the runners up are Jason Varitek and Jorge Posada, who are tied at 69 as of this writing. Although Martinez (+11.7) and Varitek (+11.9) have had good seasons, these are countered by the disappointing years of Ivan Rodriguez (-15.4), Javy Lopez (-14.2), and Jason Kendall (-14.4) among others--this in spite of the fact that PECOTA recognizes the aging problems that past-peak catchers encounter and had notably conservative projections for all of them.
A stroll through our league positional averages reports tells an interesting story. Twenty years ago or so, catchers out-hit shortstops by a fair margin; that’s actually been true for most of baseball’s history. Now the reverse is the case. Some of that is the presence of players like Derek Jeter and Miguel Tejada--guys who might have wound up at third base before Cal Ripken changed the precedent--but there’s also reason to believe that the job of being a catcher is becoming more difficult. I’m not talking about controlling the running game, a relatively minor component of today’s game. I’m talking about the catching part itself. Pitchers are throwing more pitches than they used to, fewer of those pitches are being hit into play, and the pitches are more likely to be thrown faster or with funkier movement (e.g., split-finger fastballs). Being a catcher is more taxing than it used to be, and there’s less room for error--and catchers are paying the price with their offense.
With that tangent out of the way, we should give our due to Barrett and LaRue. To think that Cub and Redleg fans thought they had no reason to watch their teams play out the schedule! LaRue’s story isn’t particularly interesting--a veteran having a modest career year in a down season for the position. Barrett, on the other hand, is one of those players that might “break” PECOTA. As I mentioned in Barrett’s player comment in this year’s annual, while there’s generally nothing important that happens when a player switches organizations, there may be some exceptions. Barrett escaped a toxic environment in Montreal and a manager who didn’t believe in him for a notably more stable milieu in Chicago, and has responded by finally living up to his minor-league potential.
Closser was a worthwhile experiment in Colorado, but it’s turned out terribly; a .220/.316/.375 batting line for a catcher with marginal defensive abilities would be problematic in Shea Stadium, let alone Coors Field. He’s also one of those guys that had a ton of variation in his PECOTA forecast, as is typical for players that reach the majors a bit late--his 90th percentile EqA was .316, and his 10th percentile EqA, .211. As barren as the Rockies’ organization is, they can probably afford to give him another 250 PA, but they’ll need to have a better contingency plan than Danny Ardoin in place.
Beat PECOTA: Derrek Lee, Cubs (+61.0)
Beaten Down: Jim Thome, Phillies (-55.8).
Non-Injury Division: Hee Seop Choi, Dodgers (-33.2).
PECOTA thought very favorably of Lee, giving him a 21.6% breakout rate, though of course his production this season has exceeded even its wildest expectations. The lesson here is that if you’re looking for mid-career breakout candidates, the combination of good secondary attributes (speed, defense, body type) and a very clean bill of health is a powerful one. Lee fits this profile perfectly; so do Andruw Jones and Alex Rodriguez. It’s no coincidence that the one time BP really kicked some butt in Tout Wars NL is the year the draft was held in Chicago, and Will Carroll and I got to team up.
I won’t say much about Jim Thome, other than that the Phillies are going to have one hell of an interesting situation on their hands next spring. Choi’s numbers have not been terrible, and the foul there probably belongs with PECOTA, which anticipated a huge breakout. Subjectively speaking, given how Choi’s numbers tanked after Dusty Baker gave up on him in Chicago, it’s tempting to think about how he might respond from playing most every day, rather than being jerked in and out of the lineup and shuffled from one club to another. On the other hand, he’ll be 27 next year, teams have a right to be demanding of their first basemen, and confidence is a skill.
Beat PECOTA: Brian Roberts, Orioles (+42.7)
Beaten Down: Mark Bellhorn, Yankees (-35.2)
A volatile year for second basemen; besides Roberts and Bellhorn, Mark Ellis (+25.2), Chase Utley (+24.0), and Jorge Cantu (+23.6) all did big things, while Bret Boone (-31.7), D’Angelo Jimenez (-29.9), Junior Spivey (-29.3) and Mark Loretta (-26.9 on a notably pessimistic PECOTA) all joined Bellhorn in the tank. Among the breakouts, Utley looks like the best bet going forward; he’s younger than Roberts and Ellis, and doesn’t have Cantu’s plate discipline issues. You have to hate to see the injury to Roberts, particularly as this season was mostly a power breakout and the injury was to his left elbow, which is going to take a ton of torque for a guy who takes most of his plate appearances from that side of the plate.
Bellhorn has fallen into something of a Saberhagen Oscillation, and it says something that the two teams that are probably best equipped to appreciate his skills have punted him. One thing we should look at in the future is whether extreme “Three True Outcomes” players are more-slump prone. I will say this: guys who work extremely deep into the count don’t leave themselves a whole lot of margin for error if there’s just the slightest tick in their timing. This year, Bellhorn is hitting just .098 with two strikes on him, with strikeouts in 55% of those plate appearances.
Beat PECOTA: Morgan Ensberg, Astros (+36.3)
Beaten Down: Scott Rolen, Cardinals (-57.1).
Non-Injury Division: Mike Lowell, Marlins (-34.4).
PECOTA predicted something of a rebound for Ensberg, but he’s still managed to blow away his projection. The sudden disappearance and reappearance of his isolated power while his other statistics remained just about the same, looks like it might be an unreported injury. Actually, the other statistic that has fit the pattern are Ensberg’s fielding numbers: his FRAA was +9 in 2003, -13 in 2004, and now it's back to +9 in 2005. Third base is a throwing position; it’d be interesting to raid the Minute Maid Park training room and take a look at some shoulder X-rays.
Speaking of shoulder injuries, take a look at how much Scott Rolen struggled as he tried to play through his problems. Mike Lowell doesn’t have any such excuses, and has played just about as poorly as you can without getting benched. Although I suggested back at the trade deadline that this might be the beginning of the end for Lowell, I’m not so sure upon closer examination. All the underlying metrics remain the same: his plate discipline, groundball/flyball numbers and speed metrics are about what they’ve always been. His defense continues to rate well, and he’s hit plenty of doubles. It’s just that 20 home runs seem to have evaporated; I expect that he’ll get about two-thirds of those back next season.
Beat PECOTA: Michael Young, Rangers (+50.6)
Beaten Down: Cristian Guzman, Nationals (-27.2)
PECOTA just completely whiffed on Young, giving him a 37% collapse rate, and remaining skeptical enough of his power that folks like Mark Grudzielanek showed up as comparables. Nor does Ameriquest Field provide much of an explanation; Young is one of the few Rangers who has hit just about as well on the road as he has at home. His plate approach wouldn’t appear to have changed that much, and I have no real answer here, other than the handy "ballplayers are funny."
That Guzman managed to win his division in spite of a projected batting line of .263/.301/.367 is remarkable; he comes in just ahead of the man who replaced him in Minnesota, Jason Bartlett (-26.8). Among Guzman’s top five PECOTA comparables, three were Larry Bowa, Ozzie Guillen, and Red Schoendienst, so there’s a good chance that he’ll continue to annoy us for years to come, even if his batting skills have atrophied.
Beat PECOTA: Jason Bay, Pirates (+41.0)
Beaten Down: Barry Bonds, Giants (-61.0).
Non-Injury Division: Eric Byrnes, Various (-31.1).
My pet theory on Jason Bay, who is developing into one of those multitalented players that PECOTA should treat very favorably going forward, is that players from Canada develop slower than the norm. Larry Walker? His best seasons according to our WARP system came at ages 30 and 34. Matt Stairs? Not established in the big leagues until 29. Corey Koskie? Stuck in the minors until 26. Tip O’Neill? His batting average jumped by more than 100 points at age 29. Terry Puhl peaked early, I guess. There is some rationale behind this; hockey weather prevails for about three-quarters of the year in Canada, and so players from our friendly neighbor don’t get as much experience under their belt as amateurs. Bay played his college ball at Gonzaga and something called North Idaho College (thank you, Pirates Media Guide), which isn’t much better.
Surfer boy Byrnes, on the other hand, who grew up in Redwood City and went to college at UCLA, became utterly lost once he found himself beached thousands of miles from the Pacific. He’s hit .280 in the California Republic this season, and .204 everywhere else.
Beat PECOTA: Brady Clark, Brewers (+33.8)
Beaten Down: Corey Patterson, Des Moines (-39.5)
I’m not really buying the Brady Clark breakout, which is largely batting average and playing time driven, but to his credit he’s developed into a useful complementary part now that the Brewers have discovered that he can play a credible center field, seeing a lot of pitches for a guy that doesn’t hit for much power and taking more than his fair share of them for the team (16 HBP). He’s also 10-for-23 on stolen base attempts; look for a challenge trade that brings the “brooding” Aaron Rowand (-15.1) to Miller Park next season.
Patterson started taking more pitches toward the end of last season, to the extent that I thought he was a real breakout candidate and told friends that I wouldn’t trade him straight-up for Andruw Jones (+15.6), even before salary considerations. The Cubs are probably right that he doesn’t have a future in Chicago, although that leaves open the question of just how responsible they are for his demise. I’m not talking about "teaching plate discipline," which is probably impossible once a player has hit age 25 or so. But to play Wednesday Morning Armchair Psychologist, it would seem that the Dusty Baker tough love approach is incongruous with a moody player like Patterson; Patterson certainly didn’t respond well after his demotion to Iowa. Next year, we’ll get Felix Pie up here, and the cycle shall repeat itself.
Beat PECOTA: Emil Brown, Royals (+28.4); Geoff Jenkins, Brewers (+28.4).
Beaten Down: Sammy Sosa, Orioles (-33.1).
There’s nothing all that interesting about Jenkins’ season. Like J.D. Closser, Jenkins had a relatively volatile projection, with breakout and collapse rates both around 20%, the latter because an injury-prone corner outfielder having a down season as Jenkins did in 2004 can sometimes be a bad sign (think Candy Maldonado). He’s managed to stay mostly healthy and perform toward the better end of what he’s capable of; his .292/.371/.512 batting line is an almost perfect analog for PECOTA’s .294/.372/.530 90th percentile projection. As for Emil Brown--we thought so much of him that we didn’t even print his PECOTA card prior to the season. I have one stored on my Toshiba; his top comparables are: Chris Jones, John Wehner, Tom Paciorek, Thomas Howard, and Don Taussig. If you could see this coming, please drop me a line; I’ll be happy to set you up with my Party Poker username and password.
Zen koan for the day: if a monk comes up to you and wagers you 100 lotus flowers that Sammy Sosa won’t make the Hall of Fame on the first ballot, do you accept it? Sosa’s comparables were about evenly divided between players like Reggie Jackson and Dave Winfield that managed to provide some value in their late thirties, and others like Gil Hodges and Dale Murphy who fell apart completely. What you don’t see is guys who had a rough year at age 36 and then rebounded. One change I’d like to make to PECOTA this winter is to rig the system such that the most recent season is given relatively more weight in the case of very old and very young players, and relatively less weight in the cases of player in mid-career. Before you ask: yes, there is a statistical basis for this. If Pat Burrell or Adrian Beltre has an out-of-character season, I expect a lot of regression to the mean; if Sosa or Moises Alou (+30.6) does, that’s something different.