Working the Numbers: the AL/NL transition

Everyone knows that there’s a talent divide between the American and National Leagues, but it’s really hard to quantitate. How much can we expect a pitcher’s numbers to change when he changes leagues? In an attempt to answer that, I took pitchers from the past four seasons who threw at least 100 IP in both leagues either in the same year or consecutive seasons. These criteria resulted in only 19 hits, and CC Sabathia shows up three times in there, so I’m not sure about how reliable these data are. But here are the translations as applied to some interesting case studies from this year:

Correlation compared with traditional MLEs

Dan Szymborski has posted over 30 years of minor league MLEs (hat tip Erik Manning at FanGraphs). That’s a crazy amount of work. Here are some of the players Manning highlights from 2009 and what our projections have to say about them:

Ruben Gotay, 2B (AAA)
zMLE: .258/.390/.402
Correlation: .257/.374/.423

Ready for the Big Time: Correlation MLEs for pitchers

If you have already read my article on hitting projections, you’ll understand that we apply the exact same principle to pitchers. This time I looked at pitchers who threw at least 60 innings at Double-A or Triple-A from 2006-2008, then matched that innings total in the Majors the following year. I identified 54 pitchers who went from Double-A to the Majors, and 108 pitchers who made the jump from Triple-A in that period. I monitored key rate stats, such as IP/G, K/9, BB/9, H/9 and HR/9, and came up with equations that would “translate” a minor league performance to the bigs, regressed against all those guys. That gives us everything but ERA, which is just calculated using FIP. And here’s the result for some interesting rookies this year:

Ready for the Big Time: MLEs by correlation

With the 2009 season wrapped up at the Double-A and Triple-A levels, I’ve already started working on my projections for next year. I really wanted to improve my minor league projections for this iteration, so I sat down to think about how to achieve that.

Linkage: Run values in and around the strike zone

For those of you who are interested in PITCHf/x, I found this post by Max Marchi to be really fascinating. The short of it is that it’s better to throw strikes than balls, and if you can paint the corners, all the better. But you can see some of the platoon splits on fastballs vs curves vs sliders, and it’s not at all what I expected. Pitches tend to be more effective high in the zone, not low.

Also, HITf/x is now publishing data! Here are the batters who punish the ball the most on average, for the month of April. Also, it turns out that batters hit stuff high away much harder than when they get jammed inside. Cool stuff.

Analyze This: Cliff Lee schools all doubters, AL

I’ll admit it. Even after his 2008 Cy Young season, I wasn’t completely sold on the breakthrough of Cliff Lee. Oh, last year was last year, and we all rationalized that Lee benefited from a great 78.3% strand rate and a middling .305 BABIP against. He’ll walk more guys and run into more bad luck next season; those innings are gonna wear on him. So I pegged him for 11 wins and 180 IP this season, posting something like a 3.76 ERA, 1.26 WHIP and a 6.59 K/9. The thing is, some of that has normalized and Lee is still on track to surpass basically all of those marks this season, and he just pitched another gem against division rival Chicago. Let’s see what we can learn from this year’s numbers:

Linkage: PITCHf/x tool @ BrooksBaseball.net

Hey, if you haven’t checked it out yet, click here for the PITCHf/x tool at Daniel Brooks’ site, BrooksBaseball.net. Great tool that allows you to look at any pitcher’s performance on a game-by-game basis. Also break down by balls and strikes, pitch type, at-bat, and opposing hitter. You get stats based on pitch type (velocity, break, percent strikes, etc.) plus graphs of pitch trajectory, speed, location, release point, spin and movement. It also gives info on pitch sequence, which is nice. There’s even a normalized strike zone map for the game, in case you want to check out that questionable called strike which you KNOW was a ball.

Great tool, and if you want prettier pictures, you can download data in spreadsheet format.

Ready for the Big Time: Matt Wieters

You all know his name by now, the 6′5″ behemoth known as Matt Wieters. Not only was he hugely touted as a top catching prospect when the Orioles drafted him 5th overall in 2007, but he made good on that promise by turning in a .355/.454/.600 line at High-A and Double-A in 530 total PA at just 22 years old. No, you’re not misreading that line, and I didn’t leave off the batting average. The Orioles have started him off at Triple-A, but they figure to bring him up by June if he keeps that up. Projections are all over the map for Wieters (understandably), who is already this year’s hyped prospect. PECOTA is notably bullish on him, projecting him at .311/.395/.546 in 370 PA, while CHONE sees him as a .274/.352/.439 performer.

Linkage: Baseball-Reference.com sporting a new look

In case you haven’t noticed yet, Sean Foreman has unveiled a major update to his already great resource, Baseball-Reference.com. The new format makes for much better readability (in my browser, anyway) and has new export features so you can get the CSV or the good old preformatted text versions. Minor league player cards immediately give you stat splits at each level, which is great. Another plus, Caught stealing is now included for players.

If you want to get a handle on all the new features, check out this flash tutorial.

Awesome.

Working the Numbers: On BABIP estimation

Most of us statheads have read this article by Chris Dutton and Peter Bendix already on improving BABIP estimation. The longstanding shortcut has been to take line drive percentage and add .120 to that number, but this excellent article blends in speed, batter eye, park factors, etc. Clearly, we want a better estimate of BABIP than the old addition method affords, but most of us don’t want to have to calculate all these other secondary stats (hitter eye, spray factor, speed score) in order to get xBABIP. Is there an easier way to get a better estimate?