12th December 2005, 11:00 AM
|
|
Member
|
|
Join Date: Jan 1970
Location: Sydney
Posts: 402
|
|
Quote:
|
Originally Posted by Sportz
Yeah, well you would do that if you wanted to make it easier to understand. And you would also make it clear what that $ value actually represents. The horse's official starting price or it's TAB price or what?
|
I suspect such details aren't terribly important, because alternative groupings should give similar results.
However for the purists:
Column 1 groups (final) NSW divs in first whole $ clusters, then $10 clusters as samples diminish.
Column 2 is sum(AUS Divs)/sum(NSW Divs) expressed as a +/- %.
Note that I sum the divs before taking the ratio as I believe that minimises distortions.
If I instead summed the ratios first and then added, weirdos could give problems.
e.g. the highest ratio is an amazing 4.705 = $36.70/$7.80. How much would that distort a column with an otherwise ratio of ~1.20? |