Thanks, Marshall, this puts me on the right track. When I uncomment so as to see %best4, I see what I'd want to look at to compare data. I think my number one's have too much weight on them now. It wouldn't be hard to make the probabilities closer to a coin flip.

In order to follow up on this, I'd have to do some comparing with other years. So far, I've only come upon one other bracket: wayback machine for 2006 bracket

I might have more time for this, but I'm still hooping it up, trying to reclaim my former glory.

Directory: my_data/06-04-2019-14-41-34
1.gonzaga      10
1.va      8
1.nc      7
2.ky      6
{
  "1.duke"      => 6,
  "1.gonzaga"   => 10,
  "1.nc"        => 7,
  "1.va"        => 8,
  "2.ky"        => 6,
  "2.mi"        => 2,
  "2.miST"      => 4,
  "2.tn"        => 1,
  "3.houston"   => 1,
  "3.lsu"       => 3,
  "3.purdue"    => 2,
  "3.texTech"   => 1,
  "4.flaST"     => 1,
  "4.ksST"      => 1,
  "4.vaTech"    => 1,
  "5.marquette" => 1,
  "5.msST"      => 1,
  "6.iowaST"    => 1,
  "6.nova"      => 1,
  "7.cincy"     => 1,
  "9.ok"        => 1,
}
$
[download]

Auburn versus Virginia just began, so we have it queueing, and I will live in a world bereft of sports news until I get to see it. DVR and perl are great things that happened in the digital world....