How to determine which SP is "better" than his opponent

[Professor1215]

Would like to hear some opinions on this topic. I have been experimenting with this for close to 6 weeks now and I have yet to find anything solid.

I have been using team specific stats vs all active batters via baseball reference. What I am finding is that 80% of the sample sizes are too small. I use AVG/OPS/KBB and compare each SP. Stats are skewed.

Park specific ERA won't work either in theory because those historic results contain batting data from players not playing for that team.

I am stuck here. Has anyone found anything that seems to be a little more stable? Right now all I can tell you is that KBB is the most highly correlated pitching stat with winning percentage. Everything else seems like a crap-shoot to me, but I know there is something out there.

Thanks

[Professor1215]

Today would be a perfect example of my small sample size. Kevin Millwood sported a .339/.980/1.30 in 289 AB vs current Yankees. That is actually one of the higher sample sizes I have. Based on those statistics, the fact that the Yankees had a much stronger lineup AND were playing in a hitters park led me to believe he was going to get smacked around.

His line today was 7 IP 3 H 4 BB 1 ER 6 K.

Good thing I was tied up with the family all day or I would have gotten cracked...