Everyone who has followed the Astros knows that pitching has been a key to separating Houston's team from much worse losing ways, like those enjoyed (well, maybe not really enjoyed) by a 100 loss team like the Pirates. The Astros have struggled mightily on offense--probably the worst Astros' offense of the decade. But the pitching seems to have held things together, so that the team has a chance to win a lot of one or two run margin games. Remember that, as the season began, the statistical projections for this pitching staff were not very good.
Some sabermetric-oriented fans express surprise when they stumble across the Astros' team pitching stats. For example, the Astros' starting pitchers posted the 6th best ERA (3.78) in the NL. The "Astros are a terrible team" refrain has become so ingrained that some bloggers assume it is a fluke. And, frankly, anytime we evaluate pitching success, we always have to be worried about illusory pitching success. On the offensive side, we worry about regression for high BABIP hitters like Chris Johnson. But what about the pitching side? Is that a bigger concern for regression next year?
I can't answer the question definitively. But I think we can look at some team level leading indicator type stats to suggest whether that is a concern. Certain indicators that pitching success is caused by luck rather than true skill are frequently used to evaluate the potential for individual pitchers' regression. I'll look at some of those indicators at the team level.
This stands for Fielding Independent Pitching (FIP) with HR per Flyballs normalized. FIP is better than ERA at projecting future pitcher performance, and pitchers' HR / flyball rates tend to regress toward the league mean. Pitchers' primary control of HR rates is connected to their groundball and flyball rates. This isn't to say that pitching skill has no influence on HR per flyball rates, but the ability to keep flyballs in the park involves quite a bit of luck, since the difference between a HR and a ball caught at the wall, or a double off the wall, can be a matter of inches. Generally x-FIP is a good predictor for the future. That's good for the Astros, because the Astros have the 5th best x-FIP for starting pitchers in the NL.
Difference Between ERA and FIP
As mentioned above, FIP is a better forecast for future pitching performance than ERA. An ERA which is better than the FIP is usually considered a sign that the ERA is likely to deteriorate in the future. So, a negative difference between ERA and FIP (i.e., ERA is better than FIP) is a bad indicator, and a positive difference is a good indicator. The Astros starters have a higher ERA than FIP (3.86 ERA vs. 3.78 FIP). That, too, is a good sign for the future. At the least, it doesn't give us cause to believe that the ERA has been flukey.
Batting average on balls in play (BABIP)--ah, you know we talk about this when we worry that Chris Johnson may be overperforming. But BABIP probably is even more telling for pitchers, since DIPS theory tells us that pitchers have limited control over BABIP. The Astros' starting pitcher BABIP is .304, which is right around average. The Astros rank exactly at the median BABIP (8th) among NL teams. BABIP tells us something about pitchers' luck. And it also tells us something about team defense, since BABIP is the inverse of team DER (defense efficiency ratio). The Astros' pre-season design was to have a better than average team defense, but it didn't work out that way. The Astros team defense was not very good. BABIP is also influenced by groundball and flyball ratios, since groundballs create more hits and flyballs create more outs. The Astros' starters were a tad above average ( ranked 7th), but not by much, in GB% and GB/Flyball %. So, the Astros middle ranking in BABIP doesn't seem to be driven by the type of batted balls. So, again, the Astros' BABIP seems to be what we would expect. This doesn't indicate a flukey BABIP-fueled success for the Astros. If the Astros' defense becomes worse next year, maybe we will see more pitching problems. I haven't studied that possibility in detail. But my guess is that the Astros' defense will be about the same next year. The older players may get worse on defense, but the younger players are likely to improve, which probably means the two effects will offset each other.
Left on Base Percentage (LOB)
Frequently we hear that a pitcher was lucky because he had a flukey LOB rate. This isn't really my favorite leading indicator stat, because LOB involves a lot of skill. Ace pitchers will post good (high) LOB rates year after year. Some pitchers become very good at working their way out of jams, either through mental concentration, pitch selection, ability to ramp up velocity or the types of pitches they throw. However, exceptionally high or low LOB rates probably indicate that luck has played a significant role. LOB rates are kind of like clutch hitting stats--- yes, some players show consistent clutch performance, but clutch hitting can be so far above or below the hitters' normal hitting that it has to reflect a good bit of luck. The Astros' starters have the fourth worst LOB rate in the NL. So, LOB does not indicate that the Astros' starting pitching has been exceptionally lucky. The bottom group of teams, in terms of LOB rate, tend to have young rotations. I suspect that the pitchers' lack of experience has an impact on the poor LOB rates. Watching young Astros pitchers like Paulino and Norris leaves me with the clear impression that working out of jams is a learning process. Both luck-based regression and increased experience for young starters like Norris and Paulino seem to point toward improved LOB performance.
The leading indicator stats at the team level do not provide any support for the notion that the Astros' starting pitching performance has been flukey. If anything, the indicators provide some (perhaps mild) support for predicting improved performance next year.
I should point out that examining the leading indicators at the team level may not be as effective as detailed pitcher-by-pitcher evaluations. Team level evaluations can suffer from the old saw about statistical averages, as in "he drowned in a river with an average depth of 2 feet." But I also think that a high level overview can be helpful because it depersonalizes the evaluation, so that we are not tempted to become overly enthusiastic or depressed about individual pitchers.
You will notice that I didn't review relief pitching. Two reasons for omitting them: first, I have my doubts that some of the indicators (like FIP and x-FIP) are great leading indicators for relief pitchers; and second, the bullpen tends to have more turnover from year to year than the rotation. However, if you are concerned that these indicators would predict that the bullpen will be worse next year, let me allay your fears. Generally that's not the case.
So despite the gloom and doom pre-season forecasts about the Astros' pitching, don't buy the commentary that the Astros' rotation just got lucky. I laughed at this snarky comment at Baseball Think Factory when the Astros' preseason prediction was presented:
Given that Brian Moehler is signed and should not be pitching in the major leagues, the obvious solution is to train him to enter the dreams of opposing players and murder them in their sleep.
The good news is that the Astros may have enough pitching depth next season that they don't have to consider Mr. Moehler as a staple in the rotation.