Tale of Two Pitchers

I was introduced to Google Reader about this time last year. I filled my RSS feed with sites focused on sabermetrics and my baseball wisdom soared.  I've added tons of new acronyms and abbreviations to my analytical palette but the real benefit has been shaping new perspectives on the game.  Today's post follows a new angle into pitchers that I previously hadn't considered before.  In the spirit of the playoffs, I'd like to compare two postseason pitchers. Here are their career lines.

Pitcher A: 529 IP, 3.56 ERA, 3.19 FIP, 1.27 WHIP, 714 K, 7.3 H/9, 0.9 HR/9, 4.1 BB/9, 2.99 K/BB
Pitcher B: 558 IP, 3.66 ERA, 3.78 FIP, 1.33 WHIP, 487 K, 8.7 H/9, 0.8 HR/9, 3.3 BB/9, 2.39 K/BB

On the surface, they have enjoyed pretty similar careers, no?  Well, Player A is Brad Lidge and Player B is Jon Lester.  Now you know I omitted some of their other statistics to make this a little more fun.

Their ERA and innings pitched are nearly identical but Lidge has made exactly one start his entire career and Lester has made exactly one relief appearance.  For this reason, we interpret their career lines a little differently.  Lidge has an infamously up-and-down career while Lester has been rock solid throughout.  Let's look at another comparison. 

Stretch A: 58.2 IP, 7.21 ERA, 11.0 H/9, 1.7 HR/9, 1.79 K/BB
Stretch B: 59.1 IP, 6.07 ERA, 8.2 H/9, 1.2 HR/9, 2.95 K/BB

Stretch A, as you might have guessed, describes Brad Lidge's entire 2009 season.  Stretch B covers Jon Lester's first 10 starts to the 2009 season.  Both were poor but Lester's season obviously didn't end there unlike Lidge.  Lester had another 150 innings to redeem himself as a starter.  After those ten starts, Lester consequently did his best Donald Zackary Greinke impression and catapulted himself into Red Sox #1 starter status.

Every pitcher will inevitably hit a rough patch at some point.  If the only window into Lester's year was his first 10 starts, it would be seen as a tremendous flop.  Hell, even Greinke gave up 20 runs across four starts this year.  For relief pitchers like Lidge, a rough patch can "last" a whole season.  How should we interpret that?

Those over at The Book blog discussed this very topic with the takeaway: we shouldn't lend too much credence to a reliever's inherent small sample size of work.

The reason that [Lidge] is on the post-season roster and may pitch in high leverage situations - maybe even close a game or two - is NOT because his name is Brad Lidge as opposed to Fred Bridge.  It is because he has been an outstanding reliever over the course of his career and the Phillies brass are smart enough to realize that those career stats are more predictive of how he is likely to do in the post-season than his 32 IP in 2009. - mgl

Lidge's career isn't just like it wasn't over after 2006.  Taking this though a step further, imagine if we sliced every starter's season into reliever-sized inning intervals.  How would we view them then?

The following exercise is admittedly a bit crude but for illustration purposes, it does the trick.  Lidge has had seven full seasons in the bigs so I sliced Lester's career into seven parts at 13 start intervals.  Here's the graph of each sample's ERA over that period.

Even when we partition Lester's career into seven samples, his ERA is still not as volatile as Lidge's full seasons. Lester's performance levels out around 3.50 ERA after his second stretch.  As a cancer survivor, Lester's admittedly not your normal pitcher but his and Lidge's comparable career length and success set up quite nicely for this analysis.  Perhaps a different starter or different arbitrary segments would yield more fluctuation.  

One caveat to this type of analysis is that pitching ability or "stuff" is not static.  In other words, Lidge probably is not the same pitcher he was a couple years ago.  His velocity is down a couple ticks and he very well could be hiding an injury but only he and the Phillies know that.  To that end, there's room for disagreement.  Nonetheless, the message still remains.  Every pitcher will look unpredictable in small snapshots.  It's the Pitch F/X analyst and scout's job to cancel out the inherent statistical noise of relievers and focus on the craft itself.