No matter what we learn from quantitatively analyzing past events, future events are under no obligation to conform. Anomalies pop up everywhere, formulae are constantly adjusted and, even if we get all that right, we still have to figure out what the statistic actually means.
One rubric that illustrates how number crunchers view the game is called "expect won-lost record," which issues from Bill James' Pythagorean Theorem. In its original version, take the total number of runs a clubs scores, square it, then add the square of that club's runs allowed. Divide that result into the runs scored squared, then multiply by 162 (or the number of games played). The final number gives a projection for how many games that club should win.
The expected won-lost record rests on a sensible intuition that there should be a relationship between the number of runs a club scores, the number it allows and the club's won lost record. So far, so good.
And it works a fair amount of time. For the years 2005-07, it hit within two games in 47 of 90 team seasons (52.2 percent). It also misses, quite often by a lot. Of 90 team seasons from 2005 to 2007, the formula missed by six or more wins 17 times (18.9 percent).
Based on the average club winning 81 games, we have a formula that misses by more than 7 percent in 19 percent of its instances during the last three full seasons. Your garden-variety telephone survey claims much better accuracy. The larger samples help, maybe.
Being fair, expected won-lost isn't supposed to be predictive. It's diagnostic. That's why the misses are interesting.
When a club wins a lot fewer games than it should based on its runs and runs allowed, something is wrong. Often, it's a matter of losing a big majority of one-run games
One club that habitually confounds its expected won-lost is the Cleveland Indians, who shared last place in the American League Central as of July 29 despite scoring more runs than they allowed. The Indians' expected won-lost said they should have held a winning record of 42-40. They were 37-45.
It's been suggested, most prominently by Baseball Prospectus last week, that the Indians shouldn't be sellers going to the trade deadline, in part because of their positive run differential. The assumption, apparently, is that the Indians' won-lost record will somehow catch up to their run differential. But it won't.
The reality for the Indians is that their run differential is unbalanced by 10 games this year in which they've scored nine or more runs. Such outbursts, though common in baseball, look freakish on the Indians because their offense is so lame. The Indians' OPS of .715 ranks 10th in the American League and would rank 13th in the National League.
The more telling statistic concerning the Indians is that they scored three runs or fewer 40 times, almost half of their games, winning only seven of them. When a formula says a club should have a winning record in today's baseball after scoring three or fewer half the time, we should pause.
In 2006, the Indians scored 870 runs and gave up 782, coming to an expected won-lost record of 90-72. In reality, they finished 78-84, 12 games off their expected record and 18 games out of the division lead.
It didn't make a lot of sense in 2006, except Indians fans remember well what happened: The bullpen crashed, totaling 24 saves for the entire season.
Again underperforming their expected won-lost record this year, the Indians have found a new way to do it. After putting together one of baseball's best offenses for the last dozen years or more, the Indians have stopped hitting all at once. For the Indians to reach 870 runs this year, they'd have to suddenly become one of the best scoring clubs of all time, averaging 6.5 runs per game for their last 80 games.
And that won't start happening soon, because injuries are killing this club. Designated hitter Travis Hafner and catcher Victor Martinez are out for the moment. Neither performed up to standard before his disabling. Martinez hasn't homered in his 198 at-bats, while Hafner was hitting only .217 with four homers.
Casey Blake leads the club in hitting at .281. Grady Sizemore is hitting for greater power (19 homers) and a little less average (.268). Ryan Garko and Jhonny Peralta both are in the low .240s.
As much as the Indians struggle offensively, they might still win if their pitching were healthy. Instead, they're missing Fausto Carmona and Jake Westbrook. Both were pitching pretty well, with Carmona carding a 3.10 ERA and Westbrook going 3.12.
So it falls to Cliff Lee (11-1, 2.34) and C.C. Sabathia (6-8, 3.78) to carry the pitching staff, along with rookie left hander Aaron Laffey (4-5, 3.24). But it can't even be that simple this year for the Indians.
Sabathia is in his walk year with the Indians in last place. The inevitable question arises, punctuated with loud talk that the Chicago Cubs will make a major play for Sabathia. Deal or no deal?
In contrast to last place clubs like the Reds or Kansas City, Cleveland is playing way below standard. The Indians took Boston to seven games in the American League Championship Series last year with the same club. They can win with this group. Westbrook isn't coming back this year (Tommy John surgery), but if they can bring back Carmona, Martinez and Hafner and Garko and Peralta pick it up, they can make a move.
The AL Central, however, is a tough environment for making a late charge. Chicago, Minnesota and Detroit all are in the thick, while the Indians were nine games from their nearest playoff position. The Indians are in a tough spot.
Their answer lies neither in the standings nor in the statistics, but at the bargaining tables. If the Indians can make a deal to sign Sabathia, they keep him. If they can make a trade for the right prospects, they trade him. If neither of those happens, they take their chances in the pennant race.
The plight of the Indians might be illustrated by qualitative analysis, but it's best understood in good old practical reasoning. The Indians won't make decisions based on statistical forecasts that they'll return to form and even out in the long run. The long run might take too long.
They'll decide based on value, whether that means money or talent. And that's the right way to decide.
Contact Bill Peterson: firstname.lastname@example.org