Friday, June 29, 2007

trend spotting update

last week's update contained this caveat:

when this writer set out to examine run differential as a potential predictor of trends in play going forward, it was not with the expectation of anything like this sort of regularity. both the cubs and the brewers have developed in a very orderly fashion within the parameters of this analysis to date -- and it is not the expectation here that this will continue to be so. while such periods represent the probabilities that makes this analysis worthwhile, there will also be false signals and periods of only very slight trending that will introduce a great deal of noise.

filter all that to mean: this thing has been weirdly accurate, and it won't be in the future. all it does is show you the trends in run differential, which -- as i at least began to show -- tend to keep trending once they've started.

in the light of full disclosure, maybe i should repeat in more blunt language than this that this is not a crystal ball. i can't tell you where whether the averages will converge or diverge or where the difference will be in a week or a month.

what i can (i think) tell you is that there is a (perhaps slight) probability of some things happening over some other things. that's all. i'm sure everyone knows that, but it doesn't hurt to repeat it for humility's sake.

to wit -- when last we left the cubs, i said:

the moving averages of run differental again negatively converged. such crossovers generally have preceded bouts of poor play of some 10-20 games duration.

there's nothing like going 6-0 to gut a forecast, is there? "poor play" -- check. remember, these are only probabilities we're talking about.

perhaps amazingly, scoring 34 and allowing 20 over the last six has only just eked the averages together -- indicating no trend in run differential at all, effectively, over the last 12 and 26 games. it's a picture of .500 baseball. this is not, you may imagine, the best environment for a forecasting tool that is based entirely on streakiness.

what happens now? beats me. if i had to guess, i'd still say that -- until a truly low minima has been reached, somewhere near or under (-1.50) -- the slightest balance of probability is toward the downside. that's hard to say when the club has gone 8-4 since your latest negative signal, but there you are.

from the outset of the year to the cyclical high on game 31, the club went 16-15; from the high to the game 53 low, the club went 6-16; from the low to the high in game 65, they went 8-4; since the high, they have gone 8-4.

in terms of crossover points, the cubs went 16-18 to the negative convergence at game 34; from that convergence to the positive crossover in game 55, they went 8-13; from there to the negative convergence in game 70, 6-7; since then, 6-1.

another picture of oddity is milwaukee, who is on the very hottest of hot streaks. their differential averages have diverged by 2.30 runs -- that's exceptional, but then this has been a team of very powerful streaks all year (two good separated by one bad). i last week said:

with their macd reading having approached an extreme level, a reversal -- though it may not be imminent -- would appear to be nearing.

they went 5-1, showing just how "not imminent" the reversal was. is it imminent now? beats me -- but the probability is yes.

all season long, people have claimed series after series to be "must-win" for the cubs -- in my opinion, "must-win" is up there in the pantheon of overused sports cliches with "clutch", "one game at a time" (do you have another choice?), "one-hundred-and-ten percent" (most athletes clearly don't take math), "taking it to next level" (if you had control of it, why weren't you already at the higher level?) and "i want to thank god" (who has better things to do than fool around with sports).

but on the eve of a three-game matchup at wrigley with the division leaders, this is perhaps one of those times where a sweep really could animate the chances of this team -- and getting swept end all doubt. i tried to point up yesterday just how truly improbable it is to outperform for a long time clubs that have, over some previous significant amount of time, outperformed you. but if it's going to happen, it has to start with getting lucky over some very small amount of time over which even mild randomness can play a very big role.

that small amount of time is here. good luck, cubbies.

No comments: