Subject: Re: CHODR Predictions AND Probabilities
In-reply-to: Your message of "Wed, 10 Mar 93 01:03:59 EST."
             <[log in to unmask]>
--------
 
Danzer <[log in to unmask]> writes:
 
> And now for this week's predictions, and, as promised, something new to go
> with them - probabilities.  We've listed the predictions in the normal
> fashion for those who just want "the latest line" but a summary for each
> series is included below.  The summary includes information such as the
> chance of a win for each team in each game, along with overall chances of
> winning the series.  There is alot of information that you may find fun
> and interesting.  First, a bit of mathematical explanation for those who
> are curious.
 
After extensive research here, using the sophisticated principle of "Occam's
Razor", I have developed an algorithm which also predicts the winners of
NC$$ hockey playoff games.
 
Unfortunately, it doesn't yet include probabilistic information.  Hopefully,
I can add it in the next release.
 
Perhaps the most interesting feature is that it makes EXACTLY the same
predictions for all the series as CHODR does, and requires substantially
less analysis and supercomputer time.
 
The algorithm is:
 
   1. Pick the home team.  Or pick the higher-seeded team.
 
That's all.  Wasn't that easy?  I guarantee this method will pick the
results of at least 75% of the games correctly -- go find your bookie! :-)
 
CHODR seems to work well as a method for ranking teams, but crowing about
its "success" in picking winners, when all you need is a copy of the
standings to do just as well, is misleading.
 
What you should really do is use it to set a point spread and pick around
that.  Ie, if CHODR says that Team A is favored by 3.4 to 1.4, and if
CHODR is correct, then the results of games between A and B should be
distributed (normally? I haven't worked it out) around a 2-goal difference.
 
Or compare CHODR's results in picking to RPICH and TCHCR in the few
instances where they're different for a good comparison.
 
Eric
 
---
Eric Scheirer -- Cornell U. -- (607) 277-1899 -- [log in to unmask]
 
"Small change can often be found under seat cushions." -- Lazarus Long