At what point do you realize your metric is garbage and go back to the lab?
Apparently OSU is in for a big season, and Purdue is going to be better without Swanigan
http://www.espn.com/mens-college-basketball/story/_/id/20163677/surprise-experienced-wichita-state-shockers-top-summer-bpi-update
You know I love me some metrics, and while I don't have much love for BPI, I think this particular complaint is interesting and worth meditating on.
I'm assuming you're not earth-shatteringly mad at the Shockers (maybe you are), but I'll look at the Purdue and OSU things.
Purdue: They return just shy of 80 percent of their minutes. I think Biggie was damn good, but I also wouldn't be blown away if a super seasoned team could be that good or better. I saw UW get better without Alando Tucker. MSU took a slight step back when it lost Green, plus another two starters from his last team. It also stands to reason Purdue is over-ranked because BPI might not put enough emphasis on starz and too much on returning players (thus hurting one-and-done-reliant teams). Of course, fans might also put too much emphasis on the biggest names, but who knows?
OSU: This one seemed weird. OSU probably didn't move up in coach given their metric, lost a lot of the roster, isn't bringing in five-stars. But then I read closer, and they really should emphasize this: "One factor not considered in preseason BPI 1.0 is transfers." So that projection assumes OSU is returning seven rotation players and adding a pair of four-stars, losing only Thompson and Loving. If that was the case, a jump from the mid-60s to 30 is not all that illogical.
Now all this asks a larger question of metrics-based projections, what do we want from them? Do we want them to about reflect our preconceived narratives? Because if we do, why would we have them? I suppose they did their job, providing something different and making us salty they don't match what we think (granted we can't often agree about how we think things will go). In theory they're supposed to give a reasonable guess of what's to come, but even if it's a really good guess, chances are it will look different enough from the final product that you can nitpick if you so choose. Seasons are long and weird with so many moving pieces. What makes a preseason projection particularly meaningful and not garbage?
(I guess I look at a computer ranking and when something is weird, I don't say, "Ah-HA, wrong." I ask why. And the why usually ends up being at least interesting)