Thursday, June 28, 2007

it is hard to make up five games

the cubs won again yesterday afternoon, but i'll admit that i barely watch them anymore. they've now gone 10-4 over their last fourteen -- and you'd think that would make any cub fan watch. they've been virtually the hottest team in baseball over that run.

except that they haven't quite been the hottest team in baseball. in fact, they haven't quite been the hottest team in their division. since june 12, the day the cubs' recent good fortune really started, the milwaukee brewers have gone a stunning 12-2 to actually expand their lead on the cubs from 5.5 to 7.5 games. and they've done it in more impressive style to boot -- while the cubs have scored 62 and allowed 56, the brewers' like figures are 96 and 53. i've found myself fangraphing the brewers far more than the cubs, which is i suppose the 21st-century equivalent of scoreboard watching. i missed a lot of cub games in 1998 by standing in the back of the rightfield bleachers facing the wrong way.

i said this a few times earlier this year and it bears repeating: five games is a lot farther than you think it is. and that is counterintuitively more true for long stretches of games than short ones.

remember all that five back business? the only other team in the sample to fall out five so quickly as this club and even get to .500 was the 1993 team. they were 38-38 at this point, and fell to 64-70 on september 2.

that club went 16-4 from there to september 24, going from six under .500 to six over. that's the kind of spell it's going to take for the cubs to get even close in this thing.

and that wasn't enough, of course. if you presume the current divisional setup, the '93 club would have been -- even after all that -- still 14.5 games short of the wild card phillies, and 3.5 short of the division-leading cardinals having been 9.5 back on september 2.

more comprehensively -- if you examine every other club in the nl over that 20-game stretch, what was the net change against the cubs over that run?

 September 3-24thru September 2

this is a table of the 1993 national league, with their records from september 3 to 24 broken out, sorted by their performance over that span. the cubs are highlighted blue, and all the teams that were worse than they were year-to-date through september 2 are highlighted in yellow.

when lightning struck out of the clear blue and the cubs went on a tear, they gained ground on everyone. but how much they gained had a great deal to do with how bad the object club was. five of the other six sub-.500 clubs fell back by 8.5 games or more. but the good clubs -- the ones that had been over .500 for the whole year beforehand -- continued either to win or at least tread water. as a result, even going 16-4 narrowed the gap by more than five games against just three of the seven clubs that had been above breakeven.

moreover, even this gain is illusory because it encompasses a sample designed not to make the league look normal but to make the cubs look really good. let's expand the sample by a week on either side -- going from august 26 to october 1.

 Aug 26-Oct 1thru Aug 26

now the cubs have actually lost ground to a couple teams, and gained more than five games on only one of the seven good clubs. and if you continued to expand the sample, in general you'd find that the cubs' performance against the part of the field that was better than they were going into the period including the win streak gets worse and worse -- until the number of games in the sample gets about as large as the number of games outside it.

the longer the stretch of games that the cubs have to outperform their betters, the less likely it is that they can. as i said above, they've scored 62 and allowed 56 in this last run, outperforming their pythagorean estimate by more than two wins. the karmically-inclined can consider it a bit of payback for the underperformance from april. the cubs have been exceedingly lucky to lose just two games to the brewers in these last fourteen, though the brewers are finding out the converse of this argument -- it's also hard to add five games to a lead.

anyway, the upshot is that good luck isn't made to last -- but when it's all you've got, you have to pray for more. john dooley in the comments complained yesterday that the team would have to win 20 in a row to make up ground. turns out that isn't far off -- so go sit in your church, spin a prayer wheel or stick a note in an old wall, per your preference. it can't hurt.

No comments: