January 24, 2003
That's the Chicago Way
The Anaheim Angels finished the 2001 season 41 games out of first place, so you would be forgiven if their World Series victory last fall surprised you. It surprised all of us. All of us except Phil Rogers, that is. He saw it coming.
He saw it coming even though last spring the Angels brought back their 2001 club nearly intact--their most significant off-season pickups were two ordinary starting pitchers, Kevin Appier and Aaron Sele.
He saw it coming even when the Angels went 6-12 against their division rivals in April, after going 17-41 against them in 2001.
He saw it coming, but he didn't have the stones to call it. Instead, he pegged them for third in the West and he has been kicking himself for the last six months because of it.
According to Rogers, "It is a very good thing to go into a season with a stable of established starting pitchers coming off workhorse seasons." Such a good thing, in fact, that you can accurately predict the postseason by finding the teams that start the year with at least three starting pitchers who threw at least 190 innings in 30 or more starts the previous season. From among those teams a world champion will surely emerge.
If you have never heard of this theory and think he pulled it out of his ass, you can be forgiven for that, too. I asked around and no one at BP had heard of it. When discussing the theory Rogers sometimes uses the journalistic "we" ("we would argue that"), but it's not as if he has a corps of colleagues at his side. As best I can tell, it's all his and he made it up to support a presupposition.
Rogers recently gave something of an explication on this theory. Last year, eight teams entered the season with at least three of these 190/30 starters, and all eight made the playoffs. Therefore, the more 190/30 men you have, the better chance you have to win the title. The Angels had the most--five--and they won the World Series.
"Even in this information age," Rogers insists, "the simplest theories still work."
His theory is simple. It makes just enough sense to lull the gullible and, if you hold a scented handkerchief to your nose, pass the smell test. If a pitcher was in the rotation long enough to work 190/30, he must have stayed healthy and pitched sufficiently enough that his team didn't have to dispatch him midseason. In theory, the more of these guys you bring to camp, the more predictable your rotation is going to be. By the same reasoning, if you were to view a season in retrospect and look for rotations that had the most 190/30 pitchers, you would have an even better chance of finding a correlation with the teams that made the playoffs.
But Rogers takes his theory one step further, using it as a predictive tool for the 2003 season. This creates a host of problems.
For one, it assumes a static level of performance and sustained good health from one season to the next. And as we all know, very little is static when it comes to the health and predictability of pitchers.
Secondly, it ignores the fact that, as often as not, pitchers who attain this workhorse status become grossly overvalued. Not all workhorse innings are of the same quality. The space between Curt Schilling and Aaron Sele is so huge that it becomes almost impossible to get a bargain on a workhorse. You either overpay for an innings-eater or you pay top-dollar for an ace. No smart team would purposely construct its rotation this way.
Rogers says that, "reliable starting pitching is perhaps the biggest factor in separating contenders from pretenders." Which might be true, but it begs the question: If reliability entails quality, how do we project reliability for the current season? Reliability based solely on last year's endurance data takes little account of injury risk, bad luck, poor performance, off-season coaching changes, midseason trades, or rookie contributions. The Angels entered the season with five 190/30 starters but ended the year with only two pitchers who did well enough and stayed healthy enough to meet Rogers' threshold. They won the World Series not because of what their Opening Day starters had done in 2001, and not because they won a war of attrition, but because of what all their players did last summer, even their rookie pitchers. It was John Lackey on the mound in October, not Sele.
Rogers boasts that last year all eight teams that had at least three 190/30 starters made the playoffs. It's true, but it's meaningless. It's a sample size fluke. And yet he bases his theory on it. That's a slim reed to lean on, and we only have to go back one year to see that it can't hold his weight.
In 2001, only one of four American League playoff teams had at least three 190/30 starters when the season began. The Mariners had only one of them--Sele, incidentally--and they went on to win 116 games. The Diamondbacks won the World Series even though they were the one National League playoff team to start the year with fewer than three 190/30 pitchers.
Three teams that year had three or more 190/30 starters and failed to make the playoffs. The Dodgers had four of them, the Giants had three and missed a fourth by only three innings. The Mets missed being that year's Angels by only six innings; if Rick Reed had pitched six more innings in 2000, the Mets would have had a full Rogers rotation.
But it worked last year, and so by applying last year's findings to this year's rosters Rogers has identified the "most likely 'surprise' teams" for the next World Series: the White Sox, Blue Jays, Phillies, and Mets. These darkhorses join Atlanta and Oakland as the only teams entering the season with at least three 190/30 starters.
He's even reluctant to call them surprise teams. Under his theory, the Angels and Giants "really weren't such surprises" last year--they were predestined for the playoffs because they had three 190/30 starters. Rogers was one of 18 prognosticators to make preseason predictions for ESPN.com, and 17 of them, including Rogers, slotted the Angels third or fourth in the West. They received 12 last-place votes. Jayson Stark had them in second, but no one outside the Angels family would have believed they would win the World Series a few months later. They were such a long shot that if Rogers' theory were credible it would be revolutionary. And who said the Giants were a surprise? Twelve of the 18 ESPN guessers had the Giants winning the division, and another five had them in second. Only Joe Sheehan had them as low as third.
Writing his article for a Chicago newspaper, Rogers' hook was that the White Sox "are in excellent shape to dethrone the Twins." Rogers nominated the White Sox as a surprise team even before they traded for Bartolo Colon. He encouraged the Sox to acquire a free agent 190/30 guy because "four 30/190 starters--provided they remain healthy--should just about guarantee a playoff spot in the American League Central." His caveat about staying healthy appears nowhere else in his argument. The article is called "Numbers Favor Sox," so the special pleading is predictable.
The three 190/30 guys they had before signing Colon were Mark Buehrle, Jon Garland, and Danny Wright. Conveniently for Rogers' thesis, Garland and Wright just do qualify as 190/30 guys, but their performance has been stagnant, with ratios that fail to inspire. Even with Colon the Sox are a long shot, but if they had won the division with just those other three, it would have been apocalyptic for the rest of us.
Whither the Twins? They "could be in trouble" because they enter the season without a 190/30 starter. Rogers notes that Radke, Mays, and Milton all went backwards last year. Well, yes, Radke and Mays were injured. Mays is still an injury risk, but Radke appears to have recovered from his groin injury. They were both pitching at the end of the season and they have had several months to recuperate.
Problem is, Rogers' theory doesn't allow for rejuvenated health. Milton threw 171 innings but that's the only sense in which he "went backward." He started 29 games, posted an expected ERA under 4.00, and his strikeout-to-walk ratio and rate of home runs allowed improved significantly. Kyle Lohse pitched 181 innings and made 31 starts as a rookie, and his workload will likely increase by at least nine innings this year, but the theory doesn't give him the benefit of the doubt. From 1999 through 2001, Milton threw more than 200 innings per season, and he's likely to do it again this year. Radke threw more than 200 innings six years in a row. Last year was the first time since his rookie year that he didn't. It doesn't matter. Rogers' data field is limited to one season. He's as near-sighted as Mister Magoo on this issue, so it's no wonder he can't see how good this rotation can be.
I don't begrudge Rogers some degree of arbitrariness, but his theory draws a bright line that illustrates its shortcomings. He lumps the Giants in with teams that have only two 190/30 men. Since they fall short of the magical threesome, the Giants--and any other team just one 190/30 guy shy of the playoffs--ought to sign one of the remaining free agent workhorses. Jason Schmidt and Damian Moss are arguably San Francisco's two best pitchers, but neither qualifies as a workhorse. They each made 29 starts last year, so no matter how many innings they would have thrown, they wouldn't count towards the threesome. There are good reasons to bet against the Giants in 2003, but this isn't one of them.
As is, the Giants are just a whisker from qualifying as a Rogers playoff lock, but rather than give the fifth rotation slot to Jesse Foppert, Jerome Williams, or Kurt Ainsworth, according to Rogers they'll have a much better shot at taking the World Series if they give big bucks to someone like Jeff Suppan. Yuck.
And then we have the 2001 World Champion Diamondbacks. By the 190/30 standard, they entered the season with only two workhorses, Brian Anderson and Randy Johnson. Curt Schilling didn't count because even though he threw 210 innings in 2000, he had started only 29 games.
That year Cleveland also made the playoffs with only two workhorses, Chuck Finley and Dave Burba. Missing the cut was Colon. In 2000, Colon made his 30 starts but managed just 188 innings. The 2001 Boston Red Sox had no workhorses, but somehow won 85 games and finished just 2.5 games behind the Yankees. The previous year, Pedro Martinez had a 1.74 ERA in 217 innings but made only 29 starts, a performance that wouldn't have counted in the Rogers analysis as much as 217 innings of league-average performance over 30 starts would have.
That's right. Jon Garland and Danny Wright make the cut for 2003, but for 2001 Curt Schilling and Pedro did not. If you had nothing else to go by, this in itself would reveal the flaw in Rogers' theory. Can you imagine using this as a team-building philosophy? As a comedian said while roasting Chevy Chase, this isn't like shooting fish in a barrel; it's like looking at fish in a barrel.
Going into this season the Yankees have at least six All-Star caliber starting pitchers. Only two count as 30/190, Mike Mussina and David Wells. Roger Clemens missed the cut. He made 29 starts, but pitched a mere 180 innings. Presumably, they need Kenny Rogers too.
But back to the point. Appier and Sele entered 2002 as workhorses, yet neither qualifies as a workhorse for 2003. That's fine, though, according to Rogers, because last season they "provided manager Mike Scioscia with enough options that he had a surplus when fresh arms like John Lackey and Francisco Rodriguez arrived." Here's where the 30/190 theory's path gets tortuous.
Appier was never "surplus," never an "option." He was one of the Angels' top three starters all year. Sele kept his rotation spot until he was injured, and he made exactly one appearance after August 20, a four-inning start on the last day of the regular season. Because Appier and Sele took regular turns, Scioscia was free to lace Lackey into the rotation? If anything, their presence would have blocked Lackey if he had no other point of entry.
Rodriguez threw 5.2 innings during the regular season. It's not as if he would have been pressed into the rotation if Appier and Sele went on the DL; when Sele hit the DL, Rodriguez was used out of the pen--exactly as the Angels intended to use him. Sele pitched only once between the time Rodriguez was called up and the end of the regular season two weeks later. He wasn't around to give Scioscia options. It's unclear how Appier and Sele facilitated Scioscia's use of K-Rod or Lackey at any time.
And Rogers was at it again on Thursday in a piece for ESPN.com. His own opinion of his theory tells you what you need to know: "Of course these are just numbers. They don't prove anything."
A priori, Rogers is rigging his argument to make it look like the surplus of 190/30 men from the Angels' Opening Day roster had more to do with the championship than it really did. He's entitled to his own opinion, but not his own facts.
Keith Scherer is an author of Baseball Prospectus. You can contact him by clicking here.