and while we are the subject of samples

[Larry Hamilton]

Sample Age is a subject that has rarely if ever been discussed on this site.

Those of you who live and breathe db's will be dismayed by this thought.

At what age do the horses pp's become too old to use, assuming the horse is still running. Say there is a 12 year old running at Delaware. Do the race this animal ran at 2 mean anything statistically? At what point (age) are an animals races useless with respect to todays race?

[cj]

Further, at what point do track "class" ratings and "par" times become useless. Its great to know a par for 10f races at CD, but if they only run 5 per year, and you want a sample size of 100 to be confident, do you want to use races from 10 or 20 years ago? Is this meets $10,000 stock on hand better than last meet's? Last year's? Or maybe not as good? Where do you draw the line?

I'd rather have solid info on the last 10 races at a given distance or class that were recently run than have the last 100 spread over multiple years.

Great point Larry...

[Fastracehorse]

<At what point (age) are an animals races useless with respect to todays race?

+++++++++++++++++++++++++++++++++++++

If a 12 yo is sharp I'll use him.

Further, models for horses in general are more important for a model of an individual horse. Ala, universiality

That is to say, db's should be used to evaluate a horses' performance based on similar horses' performances. That doesn't mean your horse isn't unique, however, there are other similar horses of equivalent uniqueness in your db ( theoretically ).

IMHO of course.

fffastt

[Fastracehorse]

<Where do you draw the line?

+++++++++++++++++++++++++

Re: rare pars.

Why not scribe a circle?? If the horse's par hits the circle, then you may have something.

Often times, horses that aren't quantifiable numerically, are great over-lays.

A good ex. is S. American imports - can't quantify but some are runners - how does one know the runners?? - another exercise.

fffastt

BTW,

The circle analogy is like trying to hit the dart board. If the horse's rating hits it, it's dangerous.

[ranchwest]

Are we only talking par times or handicapping in general or what?

I think there are certain aspects of a horse being able to get a particular distance at a particular age that could remain relevant for a fairly lengthy time.

[GameTheory]

Normally, factors derived from lifetime pp's are recency-weighted, so if 10 year pp's are even in the mix, they are given an tiny tiny weight.

For pars, I go back one year, but some go back 3 years or more I gather.

[cj]

All I was really trying to say was that we never really know how good our data is, we can only make it the best we can with what we have to work with. There are no right or wrong answers, just educated guesses.

[MarylandPaul@HSH]

I guess if you were looking at a horse's lifetime PP's from a sheets-type form perspective, you could ask how long it's been since he ran within xx points of his lifetime best. If that 12yo ran his LT best as a 4yo, and has come close to it recently, then maybe his last 8 years are relevant, at least for data analysis purposes.