Aussieplayer
06-20-2002, 11:39 PM
Hi all,
Not sure if the subject heading is quite right, as I think you refer to them as "speed lines" or "pace lines." But whatever, you know what I mean, I hope!!
Somewhat in the vein of contender selection, and freeneasy's recent post, I was wondering what your favourite "pp line selection" involved? As you can see, this is central to (most) of the contender selection methods we've talked about recently, except the contender bit goes one step further and puts in a qualification (for example, top 5, or those within 10 points etc.) to reduce the field to hopefully only a few horses.
From what I have seen, there are a few different types with various variations(!) of the same:
"Best of" - examples,
Last line, best of last 2, best of last 3 and so on and so on.
You can also bring in "exceptions" etc.
"Mathematical functions" - where one prefers to take average best 2 of last 3, median of last 5 etc.
Similar - where one prefers a line that is from a race that is "similar" to today.
Good - Where one prefers to look at "good" races only.
Recent - where the emphasis is on the last couple of races, &/or the last couple of months.
Form cycle - you might use a Pizzolla style window, with various excuses that you like, or you might try & take a rating from a similar part of a previous cycle etc.
Of course, there are millions of combinations of the above, etc. etc.
An example would be one that I gave recently, of using best of last 3 (take top 3), then best of last 2 (take top 3), and top 3 from last race. Give "poll points" 9/6/4 from best of last 3 list. 6/4/2 points for best of last 2. 3/2/1 points to top 3 using last race. Max. points for a standout horse = 18
You could do this using multiple rating systems if you wanted.
Anyway, (1) what have you seen/experimented with, what do you like?
(2) On a more sillyfossical note, is there any point to all this(!) What I mean is, without using weird things like neurals or weird mathematical thingy's like muli-nomial, repressed, depressed logit, meany, medium-rare equations.....are we doomed? Why? Why not?
(3) If one has a "way" of rating past performances that one is happy with, (ie. you believe that it does reflect a horse's performance quite accurately - let's say Beyer's, or whatever your favourite rating method is)......IS the next logical step in fact to simply test various methods such as the above out?
What a tiring process without cool 'puter stuff to do it for you - especially considering the amount of combinations possible!!
Or, is it not necessarily the next step?
Cheers
AP
Not sure if the subject heading is quite right, as I think you refer to them as "speed lines" or "pace lines." But whatever, you know what I mean, I hope!!
Somewhat in the vein of contender selection, and freeneasy's recent post, I was wondering what your favourite "pp line selection" involved? As you can see, this is central to (most) of the contender selection methods we've talked about recently, except the contender bit goes one step further and puts in a qualification (for example, top 5, or those within 10 points etc.) to reduce the field to hopefully only a few horses.
From what I have seen, there are a few different types with various variations(!) of the same:
"Best of" - examples,
Last line, best of last 2, best of last 3 and so on and so on.
You can also bring in "exceptions" etc.
"Mathematical functions" - where one prefers to take average best 2 of last 3, median of last 5 etc.
Similar - where one prefers a line that is from a race that is "similar" to today.
Good - Where one prefers to look at "good" races only.
Recent - where the emphasis is on the last couple of races, &/or the last couple of months.
Form cycle - you might use a Pizzolla style window, with various excuses that you like, or you might try & take a rating from a similar part of a previous cycle etc.
Of course, there are millions of combinations of the above, etc. etc.
An example would be one that I gave recently, of using best of last 3 (take top 3), then best of last 2 (take top 3), and top 3 from last race. Give "poll points" 9/6/4 from best of last 3 list. 6/4/2 points for best of last 2. 3/2/1 points to top 3 using last race. Max. points for a standout horse = 18
You could do this using multiple rating systems if you wanted.
Anyway, (1) what have you seen/experimented with, what do you like?
(2) On a more sillyfossical note, is there any point to all this(!) What I mean is, without using weird things like neurals or weird mathematical thingy's like muli-nomial, repressed, depressed logit, meany, medium-rare equations.....are we doomed? Why? Why not?
(3) If one has a "way" of rating past performances that one is happy with, (ie. you believe that it does reflect a horse's performance quite accurately - let's say Beyer's, or whatever your favourite rating method is)......IS the next logical step in fact to simply test various methods such as the above out?
What a tiring process without cool 'puter stuff to do it for you - especially considering the amount of combinations possible!!
Or, is it not necessarily the next step?
Cheers
AP