"I suppose the most likely conclusion is that it’s early."

I would agree.

I am very confident that Matsuzaka will finish with an ERA lower than Pettitte. Much lower.

Could someone tell me what the correlation between ERA and DIPS and FIPS is?

2+ BB innings:

Pettitte
4/05 TBD 2nd 2 BB 1 ER
4/27 BOS 5th 3 BB 3 ER
5/03 TEX 1st 2 BB 0 ER

Matsuzaka
4/17 TOR 4th 3 BB 2 ER
4/27 NYY 4th 3 BB 4 ER
5/03 SEA 1st 3 BB 5 ER

There's your answer, if we must look for a simplistic answer. Those innings above account for 35% of Pettitte's walks, and 50% of Matsuzaka's. They account for 27% of Pettitte's earned runs, and 44% of Matsuzaka's.

Were Matsuzaka to have pitched in Pettitte-like fashion in those scenarios, his ERA would have been (56% + 27% = ) 83% of what it currently is, or 3.46. Coincidentally (or is it?) 3.46 is also his DIPS ERA at this point.

Well, if you look at their FIP;

Dice-K - 3.40
Pettitte - 3.75

Sooner or later Dice-K should surpass Pettitte in ERA.

Sooner or later Dice-K should surpass Pettitte in ERA.

There's not much difference in the variance of 50 IP of ERA and 50 IP of FIP. As such, there's not much reason to expect Pettitte's current K/BB/HR numbers to continue either - he used to strike out more batters, walk a few less, and give up a few more homers. From what I've seen of Pettitte, he looks good - I figure the Ks will come.

Basically, if you give up multiple base runners in a more clumped fashion than normal, your RA will be worse than predicted by secondary stats. As many have observed, Dice-K has had trouble pitching from the stretch, which would explain why baserunners have come unusually clumped and his RA has been surprisingly high. This theory is further supported by the fact that he's been working on improving this and has been much more effective in his last 2 starts.

Perhaps it's been just random variation, but I think the thesis has good support.