Tuesday, July 10, 2012


Here at the ASB we can take a quick look at how predictive tools can go awry...the example for this installment of "be careful what pattern you see in the carpet" utilizes two teams from the NL Central who were in a virtual dead heat seven weeks ago and have since gone in completely opposite directions.

The teams? Houston (22-23 on the morning of May 26th) and Pittsburgh (21-24 on that same date).

The basic data for these teams on that day would have surely favored the Astros. Despite a rough 2011,  the team was holding its own through the first couple months of '12 and was at that time posting an overall Pythagorean Winning Percentage (PWP) of .536. Houston's pitching was holding its own (3.39 ERA).

"Over Our Heads"--another great track from
Lauren (Lo-Lo, Kotomi) Hillman, coming soon
to a soundcloud near you..
The Pirates, on the other hand, were experiencing one of the most anemic offensive starts in recent memory, with a scoring average that was fewer than three runs per game (2.82 to be exact). Their PWP was just .396.

Good grief, these things are just popping up
all over the place!!
Now, conventional sabermetric wisdom would suggest that the PWP indicated that the Astros were "for real" and could maintain their current performance level, while the Pirates were playing over their heads and would likely slide down toward the bottom of the league.

Eminently reasonable? Kinda. Sorta. But it had been a bit chilly in the Northeast in the spring (though no one except those with an axe to put in the skull of global warming evidence might possibly remember that fact right now...) and a plurality of Pirate hitters (with the exception of Andrew McCutchen) were swinging their bats as if they were giant frozen popsicle sticks.

Of course, we know what happened. That's right. The Pirates got hot and scored more runs (216) from May 26th to the ASB than any team in major league baseball. Yes, more than the Yankees, Rangers, Red Sox or Angels (with their new bashy offense led by Mike Trout and a sprung-from-his-coffin Albert Pujols). They won 27 of 40 games over this time frame and moved into first pace in the NL Central. The architects of their offensive resurgence can be seen in the comparative chart at the right. (OPS gain: from .606 to .801...or, .195.)

Meanwhile, however...the Astros, just three games out of first on May 26th, suddenly discovered that the ozone layer for their pitching staff had been sucked out into outer space (even catastrophes are more catastrophic in Texas...) and they were in more trouble than the doomed moon mission Apollo 13. That 3.39 ERA over the first seven weeks of the season was followed by a big fat 5.80 in the second (and much more unlucky) seven stanzas.

Houston allowed more runs than anyone during this time frame (245, more even than the Rockies, playing in the suddenly humorless, apparently Humidor-less Coors Field). Even our pal Wandy Rodriguez regressed as part of the Astros' "blast-off in reverse." When all the dust had settled and the ASB allowed them to (temporarily) abandon ship, the star-crossed young men had lost 30 of 41 and had managed to actually get below the Cubs (yet another piece of good fortune for Theo Epstein, whose franchise had been looking like baseball's sore thumb for the first three months of the season).

So here's another example where one of sabermetrics' primary soothsaying tools leads us down a primrose path despite its empiricist patina.  Someone needs to simply study the percentages of this particular scenario--two teams with records close to .500 with divergent PWP indicators--and see exactly how often the teams wind up confounding the scenario. Our guess,for this particular scenario at least,  is that this occurs 40-50% of the time. It'd be more clear cut over a 40-45 game stretch with teams playing under .450/over .550, of course...but it's precisely this muddled middle region that we want to be able to conquer.