MLB stats: Which ones are relevant to handicapping?

[mount187]

Reviving this thread to pose a question that I was thinking about today:

Is it better to predict how a starting pitcher is going to fair based solely on their own performance (adjusted for defense), or would it be better to start with a mean league output for SP'ers and adjust that number based on how effective a SP is. Basically, if a pitcher like Lance Lynn has made 8 starts, and carries a 2.31 ERA, is it better to just adjust his 2.31 with some other projections and come up with another "output" number (our process adjusted him up to 3.30), or should you take the league mean, say 4.50 and adjust that down to like 3.80?

The reason I'm curious, is say you have a pitcher like Cliff Lee projected pre-season to have a 2.90 ERA, and after 10 starts he has a real ERA of 1.25. Is it realistic to make a prediction based on a regression to something like 2.20? Wouldn't it be more realistic to adjust him down from the mean of like 4.60 to 3.25?

I'm not really advocating doing it one way or the other, just curious to get other people's thoughts.