I read Rick Morrissey’s column in today’s Chicago Tribune, the central claim of which is that PECOTA is liable to be an inaccurate way to predict what’s going to happen in the baseball universe, because it’s unable to account for the human element of the game:
If computers ran the world, Steven Seagal probably would have won a few Oscars by now, assuming they judged him on the $2 billion his movies have earned. If computers had a way of measuring acting ability, he’d be running a martial-arts school in a strip mall.
But they don’t run the world, yet, which means we can still type in our credit card numbers online without worrying that all our money is being sucked into a fund earmarked for global dominance by a dastardly computer.
Computers have no use for heart, or least they can’t quantify it. They can’t analyze what’s inside an athlete, for example. They can’t tell you who has the heart of a lion or the backbone of an earthworm.
As it happens, I was spending some time today going back through PECOTA’s historical projected standings. You can find a complete list of these projections, going back to 2003, right here. Our worst projection ever? We forecast the Diamondbacks to go 81-81 in 2004. They wound up playing terribly in the first half of the season and punted on the second, finishing instead at 51-111. In terms of the teams we most underestimated, there is a three way tie: the 2003 Royals (83 wins versus a forecast 64), the 2004 Dodgers (93 wins versus 74) and … the 2005 White Sox (99 wins versus 80) each beat their projections by 19 wins.
Anyway, it occurred to me that if we figure out which teams have finished the most ahead of their PECOTA projections over the course of time, we can determine which teams have the most Heart. It turns out that there is no team that has beaten its PECOTA projection in each of the last five seasons (nor has any team underachieved its projection for five years running). But if we add up the wins and losses over the five-year period, we find that the Braves finished the furthest ahead of their projections, finishing a cumulative 35 games above their PECOTA baseline. They are trailed by the Angels at +33 and the White Sox at +30; these are the baseball teams that have the most Heart.
On the other side of the equation, the Broken Hearts Club is dominated by the Diamondbacks — who had that terrible year in 2004 — with the Cubs occupying the #2 slot. Put that in your latte and stir it, Northsiders!
Getting (modestly) more serious for a moment, there may be some systematic factors in the teams that tend to beat their PECOTA projections. The teams that have finished ahead of PECOTA tend to be well-run, cohesive organizations with good scouting departments and organizational cultures; these teams may have more capacity to right the ship as the season advances onward. There also may have been some bias toward underrating teams that play good defense (which PECOTA didn’t account for until 2005) or that run the bases well (which PECOTA didn’t account for until this season).
Overall, though, the PECOTA projections have tended to get more and more accurate over time. The average error in our won-loss projections since 2003 has been as follows: