Quote:
Originally Posted by cj
...I prefer the latter because to me it shows what is really happening in the race, the horses are running evenly after the 1st call. The first method gives the illusion, at least to me, that Horses B and C are running faster the last 1/4 mile than the winner, when in reality they are running the exact same speed.
The other way I look at it is like this. Horse B had to make up two lengths after the 1st call. Is that better represented as 12 points or 4 points? Twelve seems pretty extreme for me for a horse only two lengths behind. There is plenty to time to make up the gap and that is the difference for me, horses aren't racing to the first call. Pretending they are is simply not how races are run, even in a dirt sprint. As races get longer it becomes more pronounced, and even more so on synthetic and turf.
I have plenty more, but curious to hear what people think so far.
|
Quote:
Originally Posted by steveb
No opinion, thus far, but I am just out of bed, and my brain is befuddled, so trying to unravel your post.
Personally, I have always struggled to settle on a method, where section numbers are concerned
There is umpteen ways to do it, but none of them seem to have a lot of a value when checking how predictive they may be.
By predicitive, I mean as computer model factors.
Which, I guess is my main reason for asking you.
|
I suspect the only way to know which of the two methods is more predictive is to test.
Perform statistical analysis on a clean dataset (meaning outliers have been handled) for both methods. Take the coefficients generated by your statistical analysis package for the likelihood of the event you are modeling (for example winning the race) and create two models.
Both models are identical with one exception:
The first model uses pace figs generated by method A with other factors.
The other model uses pace figs generated by method B with the same other factors.
Test the performance of both models on out of sample races and (hopefully) you'll have a clear answer.
It's possible the difference between the two methods is small enough that there's no statistical significance.
But you won't know unless you test.
Quote:
Originally Posted by denniswilliams
v=d/t
t=d/v
d=v*t
Definitely agree.
If I graph your scenario I get 3 straight lines. IOW constant acceleration. Why would anyone use pace figures that do not 'show what is truly happening in the race'? Shouldn't 'figures' capture the math?
|
Personally, I like the idea of looking at numbers that reflect what I see happening in the race.
But that's a very different thing than the 'Is one more predictive than the other and does it have statistical significance?' question Steve is asking.
Imo, that (is it useful in a model) is what really matters.
-jp
.
