Quote:
Originally Posted by cj
The biggest problem with the automated stuff, in my opinion, is that you don't catch the timing errors and human data entry errors. I process the data each night and have the program generate a list of races that look fishy and should be checked, and quite often I do find problems.
The second part is when actually making the figures. If the variant for a string of races based on my projections looks like this---20, 19, 22, 18, 41, 17---nine times out of 10 the 41 had a problem of some sort. There are times it is just a bad projection but the majority are timing problems.
Here is one example I see several times a week. Equibase has begun incorporating notes into the charts when races are hand timed. They still miss some occasionally, but most are noted. But this is a recent development which I had quite a bit of input in. But, that doesn't really help those using an automated program as it is a text field that needs to be parsed and the comments tend to vary. I know when I see a race like that 41 above it could very well be hand timed and I go read the chart and watch the race to see if fractions are displayed and if they match the chart. Is BRIS or Equibase figure makers doing that? Highly unlikely IMO. And what happens with the race that is a 41? Is it lumped in with the others? Ignored? Neither is going to make a very good figure for the race and it could very well affect all the others too.
|
I've never been a big proponent of split variants, and less a fan of projected speed figs. For my money, a maker who decides his number must reflect the horse's "true ability" has overstepped the limitations of time as a measurement of performance. Just feed it to me straight , and let me come to grips with (even glaring) outliers.
But your post, in a simple, easily understood manner, lays out a great case for such adjustments.
