Even though he seems to have hit a wall recently, Michael Pineda is having an excellent rookie season. Thus far, through 22 starts, he has thrown 141 innings striking out exactly a batter per inning with a BB/9 just above 3. All of this has left him with an excellent xFIP of 3.57, 10% better than league average. Though his results are undoubtedly impressive, his stuff is perhaps more so. Most notable is his fastball (though his slider and changeup have also been excellent), which has averaged 95 MPH, the second highest average in the majors behind Justin Verlander's 95.1.
To see how he compared to rookies of recent vintage, I looked at rookie pitchers who debuted from 2002-2006, using those years as my boundaries because I wanted pitchFX data and a few years to analyze the pitchers' post-rookie performance. I took only pitchers who threw at least 100 innings in their rookie season because Pineda has already exceeded that mark and I wanted to avoid include September call-ups and pitchers who were primarily relievers. That left me with 77 pitchers. I then looked at these pitchers' performance from their sophomore season through 2010. To analyze future performance, I used two minor variations of WAR: WAR/200IP and WAR/Season. WAR/200IP is it exactly what it sounds like, simply future WAR multiplied by 200 and divided by the number of innings pitched. WAR/Season is the WAR of all seasons divided by the number of seasons they could have pitched in (regardless of whether they did pitch or not); for someone who debuted in 2003, for example, it would be their WAR from 2004-2010, divided by 7. My reason for including WAR/Season was to keep injuries as a possibility (Mark Prior, for example, rates extremely well in WAR/200IP, but not all that well in WAR/Season).
I then compared the pitchers' rookie performances to what they did afterwards. Not surprisingly, pitching well as a rookie is a good omen. In fact, most of the relative R-squared values were exactly as one would expect. For example, the R-squared values of linear regressions of K/9, K/BB, and xFIP- versus WAR/200IP were .069, .112, and .203 respectively. I also filtered out any pitchers who debuted over the age of 24 (trying to eliminate the career minor leaguers whose stuff and careers really aren't comparable to Pineda's), and got similar results.
One finding did surprise me. Using any reasonable filters I could come up with (eg: only pitchers under a given age, only pitchers with more than X career WAR, only pitchers who were better than average as rookies, etc), fastball velocity had a higher correlation with future success than any statistical measure of rookie success. For example, the R-squared value for fastball velocity versus WAR/200IP is .278 (xFIP- vs WAR/200IP was only .203). I had known that velocity is important, but it still surprised me that it would be better to pick rookie pitchers by how hard they throw than their K/BB or xFIP.
Francisco Liriano, Mark Prior, Josh Beckett, Justin Verlander, Jeremy Guthrie, Tim Lincecum, Clayton Kershaw, Joba Chamberlain, and Johnny Cueto. Pineda's in good company (and remember that his xFIP- and FBv were significantly better than my cutoff), but this list also shows the perils of relying on pitchers. Injuries felled Prior (who had one full and two partial seasons left in him), Liriano (who has never fully regained his stuff and has only had one more dominant season), and Chamberlain (who is now a good, albeit much maligned and currently rehabbing, reliever).
Another concern is that Pineda throws so many sliders (30% of his pitches so far this year). Sliders are widely regarded as a high stress pitch, and there have been a couple of studies (eg:http://www.fangraphs.com/blogs/index.php/too-many-sliders/) showing some correlation between heavy slider usage and injuries. Filtering down the 2002-2010 group to only those who threw sliders more than 25% of the time and who threw their sliders over 82 MPH (Pineda averages 84.1) left only 4 pitchers: Francisco Liriano, Pete Walker, Josh Johnson, and Rodrigo Lopez. Obviously that sample size is too small to draw any really meaningful conclusions, but both Liriano and Johnson, both of whom seem relatively comparable to Pineda, have obviously struggled with injuries (Walker did as well, going on the DL with a sore shoulder in 2003 and 2006 before getting surgery in 2006).
I was initially prompted to take a close look at Pineda when Dave Cameron suggested that the Mariners should trade him for some offense. Basically everything I could find points to Pineda being a very good pitcher for as long as he remains healthy, but that last part is the key. It would be hard to trade someone who could be the next Beckett, or Verlander, or even Josh Johnson. At the same point, the Mariners can't hit, and selling early on Pineda could be a remarkably beneficial move if he ends up getting injured like Liriano, Prior, and Chamberlain.
Overall, I still have no idea whether the Mariners should trade Pineda, but I now have a slightly better idea of what makes a good rookie pitcher.