David W. Myers - 2001-03-16

I had a pretty big idea for adding a statistical element
to the project, and it's much simpler than what I originally planned, codewise. The research needed to set this one
up is also doable now, with a lot of grunt work.

Basically we need to get a ton of top 100 lists and
figure out means and standard deviations of the
players on real ratings lists (incidentally, I've started
on this now). We can use that to add a random
element to the rankings at the point we originally
calculate the ratings for the players. And this will turn
a program that would ordinarily be deterministic into a
Monte Carlo simulation and *that's* what I've been
wanting from the very beginning. 

With a Monte Carlo simulation, run over and over and
with some code that accumulates statistics, you suddenly
get a  feel (and the ability to predict) where a player might
be drafted. Also, you can then answer questions like:
what are the odds such and such a player will fall to
team X at pick Y...and you're doing it on statistically reliable grounds.

Grayhaired.