|
|
03-20-2018, 07:19 PM
|
#31
|
Registered User
Join Date: May 2008
Location: Nebraska
Posts: 15,123
|
Those darn hillbillies pop up everywhere. When starting this project, like most projects, I figured at first it should be fairly easy. And part of it is, and that part is getting the raw numbers. The goal is to compare different tracks to establish the speed of each track.
My plan is:
All races to be used are for 4 yr olds and up and on a fast track.
I am only using certain races. If different run-ups are used only the "standard" run-up will be used. Grade 1 and 2, and Maiden Claiming races are not used. Next I looked at each day and eliminated any day that had less than 2 and more than 5 wire to wire winners. Of the remaining races, only races that the winner was within .5 lengths of the lead for the whole race will be used.
Keeneland 6f......22.65......46.00......70.59 (2014 to 2017) 49 races
Pick the method apart.
|
|
|
03-22-2018, 07:36 AM
|
#32
|
Registered User
Join Date: Mar 2018
Posts: 17
|
Couple of questions
Let me say, I am new to these forums. Actually, Ive been away from them for a number of years. While I tend to agree with the poster who wrote:
There might be better ways to handicap races than what most of us are doing, but trying to eek out (very) slightly more accurate numbers seems like a waste of time.
I do find these discussions of academic interest. Besides you never know when you might learn something. But to my questions.
First, could you explain this line in your post?
Keeneland 6f......22.65......46.00......70.59 (2014 to 2017) 49 races
Are these the average call times for 6f races at Keenland? Is 49 the sample size you have for Keenland 6f races? If so, how many tracks are you trying to model and what are their corresponding sample sizes? I ask because this segues directly into my second question. How large is your database? More specifically, how many races in your DB fit the sampling criteria you have laid out?
Thanks
|
|
|
03-22-2018, 03:05 PM
|
#33
|
Registered User
Join Date: May 2008
Location: Nebraska
Posts: 15,123
|
My data base for Keeneland includes all turf races from 2014 to 2017, and all dirt races from the fall of 2014 to 2017. The numbers you see are the average fractions for 49 6f races that meet the conditions in the above post. This is not the average for all 6f races. 22.58.....46.11.....70.96 is the average for all races.
I also have data bases for all major and some other tracks, that range from only 2017 to as far back as 2010. The data also includes the last 3 running lines for the winner. I am doing this on this forum because there are a lot of people on here that are A: smarter than I am and B: because they might get a good laugh at how i am doing this and C: learn something along the way.
If you handicap Keeneland every horse you deal with is a shipper. I am trying to set a base to compare the horses from different tracks.
|
|
|
03-23-2018, 06:57 AM
|
#34
|
Registered User
Join Date: Mar 2018
Posts: 17
|
Quote:
Originally Posted by headhawg
...trying to eek out (very) slightly more accurate numbers seems like a waste of time. Threads like these -- while great for discussion -- seem more academic than practical......
|
You are truly wise. Measuring data with a micrometer makes little sense considering it was collected with a fish net in the first place. But I do like the academic side of these discussions.
|
|
|
03-23-2018, 06:59 AM
|
#35
|
Registered User
Join Date: Mar 2018
Posts: 17
|
If you dont mind a little conversation, what kind of data files are you working with? There are several meanings to this question. First, I have a few data source questions. Are they BRIS files, ITS files or some other kind of data files? Second are these complete files or partial files? By this I mean do the files contain both past performance and results data for the same race? I ask because I know complete files would not be necessary for your project.
Additionally, could you explain your proposed statistical and/or methodological process for making cross track comparisons in a bit greater detail.
Thanks
|
|
|
03-23-2018, 08:01 AM
|
#36
|
Registered User
Join Date: Jan 2010
Posts: 2,176
|
Quote:
Originally Posted by jay68802
Those darn hillbillies pop up everywhere. When starting this project, like most projects, I figured at first it should be fairly easy. And part of it is, and that part is getting the raw numbers. The goal is to compare different tracks to establish the speed of each track.
My plan is:
All races to be used are for 4 yr olds and up and on a fast track.
I am only using certain races. If different run-ups are used only the "standard" run-up will be used. Grade 1 and 2, and Maiden Claiming races are not used. Next I looked at each day and eliminated any day that had less than 2 and more than 5 wire to wire winners. Of the remaining races, only races that the winner was within .5 lengths of the lead for the whole race will be used.
Keeneland 6f......22.65......46.00......70.59 (2014 to 2017) 49 races
Pick the method apart.
|
I've done a similar exercise for Parx over the last year year(2017) with absolutely no restrictions. Of course there are few if any Gr I & II 6f races at Parx, usually >1M.
Parx 6f.......22.47......46.09.......71.89 (2017) 180 races
Other than class, I found that the month of the year, assuming weather/track conditions affected these times the most. I gave up using fast track designations since very fast times often show up on off tracks. I still prefer a track variant somewhat in the normal range for fast, whatever that means. If I were to remove all the MSW and Maiden claimers from the data, my guess is the final times would be very close. Looking at your TT's and final fractions compared to Parx we have:
Kee TT 23.35 ......FF 24.39
Prx TT 23.62 ......FF 25.80
The quicker FF's at Kee, most likely due to class adv. over Parx and usually better weather, but I'm not as familiar with either at Kee.
__________________
One of the downsides of the Internet is that it allows like-minded people to form communities, and sometimes those communities are stupid.
|
|
|
03-23-2018, 08:43 AM
|
#37
|
Join Date: Aug 2016
Location: Fergus,ON
Posts: 3,725
|
Comparing Different Tracks
Maybe this could help. These are the record times for Parx. For example. You might have to go through and search each track and sub out PRX in the link and maybe put SA for Santa Anita and GP for Gulfstream Park.
https://www.equibase.com/premium/eqb...trk=PRX&cy=USA
__________________
Handicapping the world year round'
-Conley
|
|
|
03-23-2018, 01:18 PM
|
#38
|
Registered User
Join Date: May 2008
Location: Nebraska
Posts: 15,123
|
Quote:
Originally Posted by LongShot_Louie
If you dont mind a little conversation, what kind of data files are you working with? There are several meanings to this question. First, I have a few data source questions. Are they BRIS files, ITS files or some other kind of data files? Second are these complete files or partial files? By this I mean do the files contain both past performance and results data for the same race? I ask because I know complete files would not be necessary for your project.
Additionally, could you explain your proposed statistical and/or methodological process for making cross track comparisons in a bit greater detail.
Thanks
|
All information is from result charts. I can get most of the information i want from there. I also have the BRIS speed and pace figures for of the winner for most races.
I do want to point out that I tend to agree with people when they say that "exact" or fine numbers is stretching or dreaming. With the way races are timed , it can not be done. I think I ended up with 134 different separations for age, class, sex, run up, distance, and conditions at Delta Downs for a 5 yr period. Add in Fairgrounds and Evd and it got to the point I said "this is not the way to go".
I will also say that in a couple areas, that includes comparing tracks to each other. It is worth the effort to be as accurate as possible.
As far as the method, KISS.
|
|
|
03-23-2018, 02:01 PM
|
#39
|
crusty old guy
Join Date: Aug 2003
Location: Snarkytown USA
Posts: 3,920
|
Quote:
Originally Posted by LongShot_Louie
You are truly wise. Measuring data with a micrometer makes little sense considering it was collected with a fish net in the first place. But I do like the academic side of these discussions.
|
Thanks LL. I've been called lots of things here but you're the first to say wise.
My point was really asking, where does the attempt at more precision stop? Let's take calculating figures to the extreme. What are the factors involved? Well horses' times, of course. To that I say, how accurate is the timing? Are the people looking for precision hand-timing the races? How are beaten lengths being determined? From charts? From watching videos and doing it yourself? What about wind? Did you account for the variability in wind speed? Surely it's not constant. How was it measured? I know this is a harness term, but what about cover? There has to be some effect on time/speed running behind other horses versus running "uncovered".
What about track surface? Banking/radius of the turns? Adjusting for differences in sea level. The exact path of the horse's trip. The alpha horse. The herding instinct...Stop me when I reach the "number of ridiculous factors" level. Measuring all of this stuff is simply fraught with error, so imo, just use the goodenough numbers.
|
|
|
03-23-2018, 07:47 PM
|
#40
|
Registered User
Join Date: Jan 2008
Location: Melbourne Australia
Posts: 918
|
Quote:
Originally Posted by headhawg
Thanks LL. I've been called lots of things here but you're the first to say wise.
My point was really asking, where does the attempt at more precision stop? Let's take calculating figures to the extreme. What are the factors involved? Well horses' times, of course. To that I say, how accurate is the timing? Are the people looking for precision hand-timing the races? How are beaten lengths being determined? From charts? From watching videos and doing it yourself? What about wind? Did you account for the variability in wind speed? Surely it's not constant. How was it measured? I know this is a harness term, but what about cover? There has to be some effect on time/speed running behind other horses versus running "uncovered".
What about track surface? Banking/radius of the turns? Adjusting for differences in sea level. The exact path of the horse's trip. The alpha horse. The herding instinct...Stop me when I reach the "number of ridiculous factors" level. Measuring all of this stuff is simply fraught with error, so imo, just use the goodenough numbers.
|
al that stuff is correct, but you are missing one very important point.
the only thing that really matters is thet you are less wrong than your opponents(others betting into the pools on the races your are)
being the modest person i am, one of the very biggest fishes, once told me that i was probably the best times guy in the world(past tense...too old and disintersted these days), but i know my stuff has lots of less than perfect stuff in it.
but it's just less wrong than most others and that's all that matters.
if more precision, even if still wrong, but wins more money than it costs to implement, then why not?
|
|
|
01-05-2019, 09:16 PM
|
#41
|
Registered User
Join Date: Jan 2006
Posts: 102
|
About 7.5 turf at gulfstream park
Would anybody post the 10k pars for Gulfstream Park about 7.5 turf.
Thanks
|
|
|
01-06-2019, 02:14 PM
|
#42
|
Registered User
Join Date: May 2009
Posts: 770
|
Does anyone happen to have a copy of the "Normative Time Table ". I believe it helps in comparing track to track ratings. I originally saw it in the book " Winning at the Track" by Mr. Christopher. I lost mine. Thanks
|
|
|
|
|
Posting Rules
|
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
HTML code is Off
|
|
|
|
|