"Rowe, Thomas" wrote:

> I don't doubt you - but the numbers on Mercyhurst and Wayne State are so high I wonder if they aren't jokes.  I suspect once you reach a certain level (like 100:1 or maybe 200:1) this all becomes meaningless.  Any numbers beyond a 100:1 should be just that - less than 1%.

Tom, I find both approaches useful in their own way.  As the song goes, "I'm always true to you after my fashion;  I'm always true to you in my own way."

The infinitesimal odds for bottom dwellers in the "Pat" version, computed consistently with the far higher results for leaders, suggest how much waddle room I should allow when supposing whether the spread between higher rated Teams C and F is really what the numbers show.
(Is that like estimating standard deviation?)

I think the "Tom" version is supported by the common sense realization that real-world factors Krach ignores don't let the odds fall below some di minimis level for any college game.  One percent might even be low, but we'd need to know the odds of an irregular flu outbreak,
a Colorado reporter's misquote that turns Yost fans rabid for Wayne State, whether Bill Wilkinson knows secrets about puck reflections off Buffalo's boards, and the alma mater of the ref's wife who filed for divorce yesterday.  Shouldn't those be incorporated into Krach?

Bob Griebel