Scott Olsen: Breakout Star?

Since the Willingham-Olsen trade went down last night, the hyping of Scott Olsen has taken on a life of its own. Early posts and comments talked about him as a "mid- to back end-starter. As the night went by, he improved to described as "solidly league average," "our new Opening Day starter," a "power arm," and even "the best arm in the organization."

Dave Sheinen's blog post this morning took the Olsen hype to strange, hilarious new heights, comparing him to Erik Bedard, Joe Saunders, and Oliver Perez. Behold this head-slapping passage:

At the very least, Olsen should be a league-average lefty good for 30-plus starts and 200-plus innings--a highly underappreciated asset in today's game. (Just look at how valuable Odalis Perez was for the Nats last year.) And at best, he could be a breakout star whom the Nats have under their control for the next three years.

My prediction: After an excellent spring, Olsen is the Nats' Opening Day starter next April.

Woah, baby! Breakout star! Let's calm down and look at the stats.

In the last few years, a whole new category of stats have emerged that attempt to isolate a pitcher's performance from the things outside his control. The reason these stats are so important if for guys like Scott Olsen, whose 4.20 ERA in '08 vastly overrates his actual performance.

Traditional stats like ERA are fraught with problems--ERA gives a pitcher credit he doesn't deserve for good defense, pitcher-friendly ballparks, effective relief pitching that strands a lot of runners, arbitrary rules about what constitutes an earned run vs. an unearned run, and dumb luck. Check out Dave Cameron's excellent, oft-cited essay on evaluating pitchers at USS Mariner for much more detail, if you're interested. These new stats look at the things that are in a pitcher's controls--strikeout rate, walk rate, groundball/flyball ratio, etc.--and factors out the "luck" factors to more accurately measure the quality of a pitcher's performance. For more background on defense-independent pitching stats, read this.

As I pointed out last night, Olsen's '08 ERA was helped enormously by a totally unsustainable .266 BABIP against (usually this number is around .290-.300, and the difference is luck and defense, not pitcher's skill).

So how good (or bad) was Olsen last year? Again, last night I walked through a series of the most concerning stats--his fast-plummeting velocity and strikeout rates, for instance. But to sum it all up, let's look at some of these new-fangled defense-independent pitching metrics. To get a range of methodologies and sources, I'm going to look at a number of these different stats:

• FIP (fielding independent pitching): This is Tom Tango stat which is based on HRs, HBPs, BBs, and Ks only and then massaged to look like the familiar ERA. I'm using the Fangraphs formula (Hardball Times uses a slightly different calculation to get at the same thing).
  
• xFIP (expected FIP): Similar to the above, but it adjusts for the luck factors associated with HR rates, like park factor and random fluctuation in HR per flyball rate, which can be quite large and not the result of a repeatable skill. This stat from Hardball Times.
  
• tRA* (I think this stands for "true runs allowed"): This is the most complicated methodology, but also probably the best. Created by Graham MacAree, tRA basically assigns a value for more or less every event that has been shown to be within a pitcher's control--not just HBPs, Ks, BBs, and HRs, but also groundball rate, flyball rate, and a bunch of other things. tRA* erases some additional randomness by regressing each pitcher's stat to the league mean. Here's a fairly simple essay explaining the stat, and if you really want to get into the muck on this one, try this one.

Here the stats for each of the seven main guys in the running for a rotation spot this year, plus Odalis Perez, for comparison. I'm leaving out Shairon Martis and Jordan Zimmermann because I don't think either should really be in the running for Opening Day and because neither has enough (or any, in Zimmermann's case) innings to make a useful stat comparison. But you can mentally note them as eighth and ninth options, if you wish:

	FIP	xFIP	tRA*
John Lannan	4.79	4.47	4.74
Garrett Mock	3.84	3.91	4.77
Odalis Perez	4.62	4.47	4.93
Shawn Hill	4.06	4.58	4.98
Collin Balester	5.11	5.18	5.10
Tim Redding	4.93	4.99	5.16
Jason Bergmann	5.12	5.12	5.28
Scott Olsen	5.02	5.05	5.29

Mock's 41-inning sample size should be taken with a little more of a grain of salt, but the bottom line here is that there's no stat that would have put Olsen in the top five starters for the Nationals in 2008. Of the group in the running for next year, FIP and xFIP has him fifth, behind Lannan, Mock, Hill, and Redding, and tRA* says he was flat out the worst of the bunch. (And take out that horrid last start from Balester in Philly, and he would be kicking Olsen all over the yard by every measure--you can't do that, because that start happened, but still.)

Olsen's 5.29 tRA* in '08 put him 109th out of 130 pitchers with 100 IP. (Coincidentally, he's squeezed right between Jason Bergmann and Bowden castoff Darrell Rasner on that list.)

Also, consider this: 2008 was actually a bounce-back year for Olsen. In 2007, he posted a 5.33 FIP, 4.98 xFIP, and 5.83 tRA* (don't ask me why his xFIP actually was a bit better in '07--this is why I wanted a range of sources and methods). That 5.83 tRA* was the 123rd best out of 129 pitchers with 100 IP.

And finally, one last thing to factor in here. Olsen's '08, as bad as it was, was significantly helped by the fact that he had the good fortune to face the Nationals--the league's worst offense--five times, throwing a total of 31 2/3 innings against us, more than any other opponent. And against the Nationals, he posted a 3.41 ERA (these fancy stats aren't available by opponent split), while posting a 4.34 ERA against the rest of the NL. Next year, he won't have the luxury of facing the hapless Nats for 15% of his IP.

The overrating of Olsen is a classic example of a psychological phenomenon known as "first impression bias," a phenomenon where people are strongly influenced by the first piece of information they learned about a person or thing. Most people heard of Olsen for the first time when he first came up in 2006, and he indeed had a very nice little rookie season.

Unfortunately for the Nationals, however, every other season in Olsen's career before and after that 2006 season tells a different story. He was a fringey prospect in the minors from '02-'05, and has been one of the worst starting pitchers in baseball in '07 and '08.

(NOTE: Basil notes in the comments that it's quite inaccurate to describe Olsen as a former "fringey prospect." As he says, he was a true top prospect with great strikeout ability and a plus-slider. The reason I'm down on him is because of what we've seen the last two years, as his velocity, K-rate, and other peripherals have fallen off a cliff. Even the slider isn't missing bats. Guys don't just lose 3 mph and 3 Ks per 9 out of the blue for no reason. I'm looking at the last two years and concluding that he's no longer the guy who was rated highly coming up.)

Yesterday's trade was supposedly made for the present. We sacrificed prospects to get better now. And indeed, Josh Willingham will be an offensive upgrade for us in left field. His bade defense will give back some of his value compared to, say, Willie Harris, but I acknowledge the Nationals will be better with Willingham. Maybe even 2-3 wins worth.

However, giving Perez's innings pitched to Scott Olsen will surely make us worse, maybe enough to wipe out all the improvement we get out of Willingham.

So here's my prediction: Olsen bursts into flames in spring training on the mound, which triggers a series of confrontations with teammates and Manny, and he has to open the season either suspended or in AAA. After a series of bad starts in Syracuse, Bowden brings him up, mostly to entertain his own ego. Olsen lasts 3 starts, posts an ERA over 10, and finally Bowden is forced to send him back down, but not before publicly questioning Olsen's desire and work ethic in what will be a successful effort to turn Nationals fans against their own player, a la Aaron Crow and Ryan Church.