The A's Starting Rotation: A Look at What's Stabilized

command was his calling card as a prospect, and his minor league bb/9 was just 1.8. perhaps his stuff is just so dramatically improved as a result of the new release point that he’s now able to pitch more in the zone to mlb hitters?

really though, its pretty amazing to look at his pitch movement charts from before and after he changed his arm angle.

2011:

2007:

getting a little more sink on that two seamer now, eh? no wonder his gb% is way up. not to mention the fact that his new cutter drops almost as much, and, at 88-89, probably looks a lot like the sinker coming out of his hand.

I’m starting to question whether the team is capable of winning three in a row.

Am I correct in stating that the team has neither swept nor been swept in a series yet this season?

You don’t have independent observations for a single player. All you have is one data point (e.g Brett Anderson has a fly ball ratio of X, not all Brett Andersons have a fly ball ratio of X, Y, and Z). What do you use for variability? A standard number that comes from looking at the variability of all baseball players after the same number of innings? This is beyond my statistical knowledge.

MobiusKlein wrote it flippantly enough that it makes it seem easy to do. Maybe it is, but I don’t understand how. From this type of data, I don’t see how you get a measure of variability to put on things like +/ estimates or 95% confidence intervals or whatever your favorite measure of variability is. We don’t have 20 samples of “Brett Anderson flyball ration up to the point of May 9th”. We have 1. Where does the error estimate come from? Can you break it up by start, in which case you have an n equal to number of starts? Or do you use baseball players as a whole and assume variability is the same for all pitchers? Bootstrapping would do it, but that takes a lot of analysis. Really, as a biologist who does a lot of stats, I am curious and want to know. I agree that this is important, and the use of the word “stabilize” is as arbitrary as alpha being .05, but I just want to know how to do it.

This is such an important point. The shift from “too-small sample” to “adequate sample” is gradual, not merely a matter of hitting a quota number of at-bats or innings pitched.

He is in the micro-walk range!

That said, I would guess the reason for this has to do with ERA being a first iteration of an attempt to measure pitcher performance, and the performance that is trying to be measured is as a thrower, not a fielder. 20+ years ago, the pitcher with the most wins (and/or highest win percentage) almost inevitably won the Cy Young. Now, at least, some attention is given to ERA along with wins. Say what you want about ERA not being all that well correlated with actual pitching performance, at least it is more meaningful than ‘wins’, where a pitcher might get a ‘win’ because he barfed up a ‘save’ and then his team rescued the game, or a pitcher who pitched 8 0-run innings gets denied that same win because the closer [Fuentes, I’m talking to you here] failed to convert the save but got the win, etc. etc.

So, in the context that Cy Young voters still pay attention to wins, and MVP voters still pay attention to RBI’s, it’s not too earth-shattering that a pitcher improves his ERA by booting a comebacker…

Maybe it’s because my last name is King, but I get thrown off every time you say ‘King’ or ‘king’ and not, like, K’ing, or K-ing.

If not, why are you citing a month’s worth of at-bats?

On an unrelated point, I continue to believe the A’s’ long-term need for Weeks is as a center fielder, not as a second baseman. I suppose they could B.J. Upton him, but moving around constantly in the field screwed up Upton’s first couple of big-league seasons.

If they’re calling up someone for second base, I’d favor Eric Sogard. Weeks can come up when Coco Crisp inevitably hits the DL.