I usually save these for the Roundup, but because this is so clever it gets its own post.
Implied power rankings according to point spread markets, complete with a very clever Moneyball reference.
The Real Power Rankings
published on 12/02/2011
Subscribe to:
Post Comments (Atom)
That was genius, and that was a clever Moneyball drop.
It's a good idea in that Vegas odds are fairly efficient, but past point spreads reflect yesterday's news. If you want to guess tomorrow's Google stock price, last Friday's closing price is going to be fairly close--but it's far from the best information we can use.
When I made my living betting on sports futures, I did this exercise by looking at the next week's point spreads, not the last week's. There's no need to consider how good a team the Texans are with Schaub healthy. Next week's lines are calibrated correctly.
sorry, but for us dolts, what is the clever Moneyball reference?
"50 feet of crap" is a line from the trailer to the movie.
j holz - but is there enough information in one week to have a closed system, so that the rankings converge on a proper solution? In my experience, you need a few weeks of data. Put simply, there are 32 teams to be ranked and only 16 pair-values in one week of point spreads.
Agree with the points of both JHolz and BB in general. The headache for BB's point is that a team's rating is different with its backup QB...which can mess up the convergence, and can throw a monkey wrench into connectivity.
I use a scale that generally peaks around 85 and floors around 70 to try and mirror the market...though it's possible for the extrems to earn their way post those (which has happened this year on both ends).
I have it for Houston:
81 with Schaub
78 with Leinart
75 with Yates
Those would correspond with the lines over the last three weeks given where their opponents in those games currently sit. But, if you're trying to decipher what the market things about Houston with Yates...then it's 75 and the past few weeks don't matter. I think that's part of what shot Atlanta up so high in the author's numbers. They're getting credit for laying 2.5 on the road to a 3-week composite when the current market just has them facing Yates.
Not sure if it helps the discussion or not, but here's where I have everyone at the moment on this scale for an estimate of how the market is rating teams through this morning's numbers:
87: Green Bay
85: New England
83: New Orleans
81: Pittsburgh, Baltimore, Dallas
80: NY Jets, Atlanta, San Francisco
78: Detroit
77: Cincinnati, Chicago (Hanie), Oakland, Miami, NY Giants, Philly (Young--estimated drop of Thursday), San Diego
76: Tennessee, Denver
75: Buffalo, Carolina, Tampa Bay, Houston (Yates)
74: Seattle, Washington, Minnesota
73: Arizona
72: Jacksonville, KC (Palko)
71: Cleveland
70: St. Louis
68: Indianapolis
Market's giving more respect to Palko this week than they did the last two...so that's a bit dicey. Interesting to me that Green Bay is getting a lot of respect in the lines, yet is still coasting past the number so often. Highly rated, and under-rated to this point simultaneously. Maybe that changes in December...
I posted more details in the other thread, but I've found that using 50% current week's spreads and 50% previous two weeks works fairly well and helps mitigate the backup QB problem.
Thanks Brian for the call out. Being a Colts fan, that was hard to write.