Sabermetrics

One thing that I think bears mentioning is what us stat-heads mean by a "better" stat.  It seems self-evident, but I think that a lot of "old school" vs. "new school" debates stem from the fact that the people debating misunderstand a very basic element about the other side's argument.  You ran into a bit of that here, and we run into it all the time at this site.

Old-schoolers are interested in a player's present value.  They have their fan hats on, and want to know how a player is affecting the team right now.  Want to know how good a pitcher is?  Easy.  The less runs a pitcher gives up, the better.  ERA is earned runs allowed per nine innings.  Doesn't get any better than that.  Want to know how good a run producer a hitter is?  The more runs a batter drives in, the better.  RBIs!  In this frame, they're absolutely right.  The best record of how well a pitcher prevented runs (or is preventing runs) is his ERA.  The best measure of how many runs a batter drove in is RBI.  What the hell is so difficult to understand about that?

We stat-heads think a little differently.  We want to project future value.  We have our GM hats on, and want to know who we should go after in the offseason.  We want to know what a player can be expected to do next season, the season after that, and the season after that.  ERA isn't as reliable for that because it fluctuates so much due to luck.  We can actually predict a pitcher's ERA better using last year's WHIP than we can using last year's ERA.  How did I predict that Jason Jennings was going to suck this year?  His ERA drop in '06 came without improving his K, BB or G/F numbers, indicating dumb luck was at work.  RBI's are a favorite sabermetric whipping boy because they're stubbornly reliant on people getting on base in front of you.  Why did Ryan Howard have 30 more RBI than Adam Dunn despite very similar batting lines?  (answer: Chase Utley's .410 OBP).  A lot of old-schoolers make fun of "luck" stats, saying "Why do we need to know what these guys would do in an alternate universe?"  Every season is an alternate universe.

That's a very long way of saying that WHIP is a better stat in that it does a better job of predicting future performance than does ERA.  If you want to know who you should give your Cy Young vote to, ERA is fine.  If you want to know who to trade for in your fantasy keeper league, WHIP is better.

I think you've got the gist of what WHIP means when you say, "the less trouble a pitcher is in the less runs he will give up," but I get the impression that you may be missing what the real goal of the analysis is.  

Ultimately, what we want to know is what pitcher can we depend on in the future?  If you see a pitcher with a higher ERA than would be indicated by his WHIP, that is, he's given up more runs than you might expect for the number of runners he's let on base, then you'd probably be better off with him than a pitcher who constantly gets out of jams because he allows a lot of base runners. Like you said, eventually it will come back to haunt that second guy.

As I'm typing this I see BLee's response and he's right about what he is saying, especially since it's basically what I am saying too. :)

One last thing I would add is that a pitcher's value becomes more predictable if we can take the affect of the defense out of the equation.  That's why you'll see most statheads spouting things like K/9, BB/9, and HR/9 when discussing how good a pitcher is.  On a season by season basis, those things are much more predictable than ERA or WHIP, especially for a guy that is changing teams.  So while WHIP may be a better predictor than ERA, I'll take the other 3 over WHIP any day if I'm trying to determine what sort of success a pitcher will have in the future.

But it's also important to know how well pitchers prevent hitters from advancing runners around the bases (e.g. doubles, triples, homers), because that's the other component to scoring runs.  And WHIP largely ignores that.  So from my perspective, using only WHIP to evaluate a pitcher's performance is like using only OBP to evaluate a hitter...

ERA includes information about not only getting on base, but also advancement around the bases.  So from the perspective of trying to understand what happened in the past, I much prefer ERA.  And I will say that some of us statheads are pretty interested in questions of historical performance. :)

As Slyde point out, though, ERA (and WHIP) can be vary wildly from season to season due to team defense quality, and probably more importantly to "luck", such as non-repeatable variation in how events are clustered (e.g. Matt Belisle's three grand slams allowed last season) or the frequency of "ground balls with eyes".  In fact, once you factor out strikeouts and homers, hit rates (Batting Average on Balls in Play, or BABIP) vary much more across years for individual pitchers than they vary consistently between pitchers.  

K/9, BB/9, and HR/9 (or, even better, GB%) on the other hand, are very stable measures of pitcher skill from year to year, and seem to be the best ways to understand the degree to which performance is due to skill or luck.  And you can use Tom Tango's FIP stat to estimate the ERA a pitcher will tend to get given their peripherals--that can be extremely helpful in trying evaluate the summed effects of strikeout, walk, and homer rates when, for example, you have someone like Ryan with great strikeouts but high walks..

Anyway, the point is that (like Slyde said) if you're after prediction, then peripheral-based analysis seems to be the way to go.  On the other hand, some statheads like myself are interested in documenting historical performance and value too.  And ERA does a fine job in that case, though many prefer RA because ERA is a bit too kind to groundball pitchers.  

I honestly don't pay much attention to WHIP because, depending on the question at hand, there are better statistics out there that are just as easily accessible.  It's not that it's a bad stat, of course.  It's just that there are better options.
-j