September 4, 2008
Looking for an Advantage
With last night's 5-4 loss to the Blue Jays, the Twins fell out of a first-place tie with the White Sox in the AL Central, leaving them a game behind with 24 to play. That one-game edge doesn't seem like much, but when you look at the various playoff predictors, you see that the systems don't give the Twins much chance to make it up, and virtually no chance to catch the Red Sox for the AL wild-card slot.
Division Wild Card Playoff Odds 32.0% 0.6% PECOTA Playoff Odds 32.1% 0.5% ELO Playoff Odds 51.9% 1.2%
Why does a one-game edge with about three weeks left translate to a nearly 2:1 advantage for the White Sox in two of the three methodologies? Well, for one thing, the White Sox have been the better team to date. Their 78-61 record is an accurate reflection of their performance in 2008; the Adjusted Standings show that their third-order record, factoring in run differential, underlying performance, and opposition, is 77-62-just one game off-largely because they've allowed about 13 fewer runs that expected.
The Twins, on the other hand, have a third-order record of 71-68, a mark that places them behind the Indians within their own division. The Twins have a whopping +70 run differential, in line with their actual record, but that differential is almost entirely driven by the fantastic job they've done hitting with runners in scoring position. Based on their statistics, the Twins project to have scored 646 runs; they're 61 runs ahead of that pace, the biggest positive gap in baseball. A team that has batted .279/.339/.410 overall has jacked that to .311/.386/.456 with runners in scoring position. The improvement shows across the board: a higher BABIP, higher isolated power, and higher unintentional walk rate.
Most research has shown that hitting in the clutch is not a skill, for individuals or for teams, over and above hitting in all situations. So while the runs and the wins created by the Twins' performance with runners in scoring position have been banked, the most likely scenario for the last 24 games is that the Twins will hit with RISP what they hit as a whole, and therefore play more like a .500 team, rather than a .550 one. Last year's team, with largely the same personnel, hit .276/.347/.412 with runners in scoring position, a slight improvement on its .264/.330/.391 performance overall. Despite the persistent idea that the Twins have a skill in this area, the most likely reason for the split is variance.
The White Sox have no such statistical gremlins in their record-they are what we think they are. The success that the fly-ball tossing starting rotation had suppressing home runs in the season's first two months has largely corrected itself. After allowing 33 long balls in April and May combined, the Sox gave up 35, 30, and 27 in the next three months. This was expected, but again, the wins banked by the results early on are theirs to keep, like the first-level prize on a game show. Offensively, the White Sox have improved on last year's disastrous, and no doubt anomalous, performance on balls in play, with a middle of the pack batting average (.265) and excellent power: second in the AL in slugging, first in homers and isolated power. Just as was the case with the Ozzieball team of '05, there's no actual sheen of smallball here, either; the Sox are 13th in the AL in steals and have laid down just 23 sac bunts all year long.
I originally thought that the schedules might be the issue, but if anything, the Twins have it easier down the stretch. Just looking at the series that the teams don't have in common or the one against each other, the Sox play the Angels, Jays, and Yankees, while the Twins get the Orioles, Rays, and Royals. Even if you figure that the Sox playing the Twins calculates as a bit easier than the Twins playing the Sox does, that's a pretty big edge for the Twins.
The ELO method of calculating the odds, developed by Nate Silver, gives the Twins the best chance of taking the division, and actually makes them a slight favorite. ELO is designed to assess a team's skill level at a given moment in time, so it may be seeing the Twins as they are at the moment, which is an improved team thanks to the sloughing off of dead weight such as Livan Hernandez, the emergence of Denard Span as an OBP guy, and a healthy Alex Casilla at second base. It may also be catching that the Sox record is still showing vestiges of a hot start driven by the low home-run rate. ELO assigns more weight to recent results, and since the end of April, both teams have identical 64-48 marks.
I tend to think ELO has it right here, not for the reasons stated above, but for this one: team performance over 24 games is largely unpredictable. The error bars on any projection for either teams' performance down the stretch swallow up any pretense of knowing what will happen, and when you consider the three head-to-head games in there, the only conclusion that can be reached is that this is a toss-up. That the White Sox have been the better team informs the discussion, but everything we know about baseball shows us that being the better team matters less in direct proportion to the number of games we want to evaluate. The White Sox could go anything from 17-7 to 10-14 over the next few weeks, and the Twins, from 16-8 through 9-15. I'm not smart enough to pick two records from those pools and tell you who will win the Central. I can say that these look to be the key factors:
Regardless of how this turns out, it's going to be a great ride. Credit both of these organizations-mostly picked third and fourth in the division-for assembling teams that could step into the void created by the failures of the Tigers and Indians and contend. The White Sox, and specifically Kenny Williams, continue to defy expectations, while the Twins are seeing the fruits of a development system that churns out pitchers like the Valley pumps out porn.