PDA

View Full Version : time analysis request for info


wesmip1
19th May 2007, 01:06 PM
All,

For those that do time analysis how many lengths do you adjust for dead, slow and heavy on average ?

Good Luck.

Chinbok
22nd May 2007, 08:50 AM
Hi Wes,

I don't rate races on heavy tracks or distances over 1800m. Here's what I use. Figures below are seconds, not lengths.

<table x:str="" style="border-collapse: collapse; width: 187pt;" border="0" cellpadding="0" cellspacing="0" width="249"><col style="width: 48pt;" width="64"> <col style="width: 69pt;" width="92"> <col style="width: 70pt;" width="93"> <tbody><tr style="height: 13.5pt;" height="18"> <td class="xl27" style="height: 13.5pt; width: 48pt;" height="18" width="64">Dist</td> <td class="xl25" style="width: 69pt;" width="92">D</td> <td class="xl27" style="border-left: medium none; width: 70pt;" width="93">S</td> </tr> <tr style="height: 12.75pt;" height="17"> <td class="xl26" style="height: 12.75pt;" x:num="" height="17">1000</td> <td class="xl23" x:num="">0.85</td> <td class="xl26" style="border-left: medium none;" x:num="">1.6</td> </tr> <tr style="height: 12.75pt;" height="17"> <td class="xl26" style="height: 12.75pt;" x:num="" height="17">1050</td> <td class="xl23" x:num="0.92500000000000004">0.925</td> <td class="xl26" style="border-left: medium none;" x:num="">1.7</td> </tr> <tr style="height: 12.75pt;" height="17"> <td class="xl26" style="height: 12.75pt;" x:num="" height="17">1100</td> <td class="xl23" x:num="">1</td> <td class="xl26" style="border-left: medium none;" x:num="">1.8</td> </tr> <tr style="height: 12.75pt;" height="17"> <td class="xl26" style="height: 12.75pt;" x:num="" height="17">1150</td> <td class="xl23" x:num="">1.05</td> <td class="xl26" style="border-left: medium none;" x:num="">1.88</td> </tr> <tr style="height: 13.5pt;" height="18"> <td class="xl26" style="height: 13.5pt;" x:num="" height="18">1200</td> <td class="xl23" x:num="">1.1</td> <td class="xl26" style="border-left: medium none;" x:num="">1.95</td> </tr> <tr style="height: 12.75pt;" height="17"> <td class="xl26" style="height: 12.75pt;" x:num="" height="17">1250</td> <td class="xl23" x:num="1.175">1.175</td> <td class="xl26" style="border-left: medium none;" x:num="">2.02</td> </tr> <tr style="height: 13.5pt;" height="18"> <td class="xl26" style="height: 13.5pt;" x:num="" height="18">1300</td> <td class="xl23" x:num="">1.25</td> <td class="xl26" style="border-left: medium none;" x:num="">2.1</td> </tr> <tr style="height: 12.75pt;" height="17"> <td class="xl26" style="height: 12.75pt;" x:num="" height="17">1350</td> <td class="xl23" x:num="1.325">1.325</td> <td class="xl26" style="border-left: medium none;" x:num="2.1749999999999998">2.175</td> </tr> <tr style="height: 12.75pt;" height="17"> <td class="xl26" style="height: 12.75pt;" x:num="" height="17">1400</td> <td class="xl23" x:num="">1.4</td> <td class="xl26" style="border-left: medium none;" x:num="">2.25</td> </tr> <tr style="height: 13.5pt;" height="18"> <td class="xl26" style="height: 13.5pt;" x:num="" height="18">1450</td> <td class="xl23" x:num="1.4750000000000001">1.475</td> <td class="xl26" style="border-left: medium none;" x:num="">2.32</td> </tr> <tr style="height: 12.75pt;" height="17"> <td class="xl26" style="height: 12.75pt;" x:num="" height="17">1500</td> <td class="xl23" x:num="">1.5</td> <td class="xl26" style="border-left: medium none;" x:num="">2.4</td> </tr> <tr style="height: 13.5pt;" height="18"> <td class="xl26" style="height: 13.5pt;" x:num="" height="18">1550</td> <td class="xl23" x:num="1.575">1.575</td> <td class="xl26" style="border-left: medium none;" x:num="">2.49</td> </tr> <tr style="height: 12.75pt;" height="17"> <td class="xl26" style="height: 12.75pt;" x:num="" height="17">1600</td> <td class="xl23" x:num="">1.65</td> <td class="xl26" style="border-left: medium none;" x:num="">2.58</td> </tr> <tr style="height: 12.75pt;" height="17"> <td class="xl26" style="height: 12.75pt;" x:num="" height="17">1650</td> <td class="xl23" x:num="">1.7</td> <td class="xl26" style="border-left: medium none;" x:num="">2.65</td> </tr> <tr style="height: 12.75pt;" height="17"> <td class="xl26" style="height: 12.75pt;" x:num="" height="17">1700</td> <td class="xl23" x:num="">1.75</td> <td class="xl26" style="border-left: medium none;" x:num="">2.72</td> </tr> <tr style="height: 12.75pt;" height="17"> <td class="xl26" style="height: 12.75pt;" x:num="" height="17">1750</td> <td class="xl23" x:num="1.825">1.825</td> <td class="xl26" style="border-left: medium none;" x:num="">2.81</td> </tr> <tr style="height: 13.5pt;" height="18"> <td class="xl28" style="height: 13.5pt;" x:num="" height="18">1800</td> <td class="xl24" x:num="">1.9</td> <td class="xl28" style="border-left: medium none;" x:num="">2.9</td> </tr> </tbody></table>

wesmip1
22nd May 2007, 06:18 PM
Chinbok,

This is much appreciated.

It is hard finding other people to compare times with as there isn't many sources of informaiton on it. Do you know of any good websites with this sought of info ?

Thanks

Chinbok
23rd May 2007, 07:48 AM
Hi Wes,

Don't know of any websites. You could try searching through the ausrace forums. Their archives go back many years.

I put my times together a couple of years ago and can't remember the exact methodology but I used 2-3 years of race results downloaded form unitab. From memeory, I think I only used the metro tracks or the metro and provincial.

Cheers

Chinbok
23rd May 2007, 07:50 AM
Wes,

Are you using track variants as well?

Racer
23rd May 2007, 10:11 AM
It is hard finding other people to compare times with as there isn't many sources of informaiton on it. Thanks
Wes.,

It's probably due to the donkeys years of research experience by such
people as the late Don Scott - His findings over many years suggested that,
Time - Pace - Speed ratings would lead one to the poor house because they are worse than useless.

His final sentence is a Gem,
" If you meet such a character, direct him to the nearest psychiatric institution. He certainly needs help."

He did like to call a Spade a Spade when it came to his passion -
Racing Research.

wesmip1
23rd May 2007, 12:27 PM
chinbok,

I have not got track variants to its full potential yet. I have noticed a few places produce very quick (or slow) times though and I do a slight adjustment for it but I would not say it is an accurate adjustment.

There are several ways to work out the track variantions and I am still condering the best way to do it. The options I have include:

1. Avg time of winner over a large number of races (500+ races).
2. Avg time of a consistent horse that has run across multiple tracks (this is good but hard to do ).
3. Track records for each distance.

Can I ask which one you prefer to use ? Depending on which one you use you are always going to get those slight differences which could cause different selections. In some races 0.1 secs can be a huge difference to whether a horse is rated as top selection or 5th selection.

Good Luck.

wesmip1
23rd May 2007, 12:31 PM
Racer,

I like the research and try and find things which work well.

Its obvious that the horse that runs the fastest time will win and therefore the horse should be rated by the potential times they could run in the race. For example a horse running 1.00 consistently should be a horse that is consistently running 1.10 .

The problem arises when horses do not run to their best times or even their avg times. This is where other form analysis needs to be taken into account to determine the fitness of the horse and whether any improvement can/should be expected.

I think the main reason people don't use it is because it is hard to find and compute yourself.

Good Luck.

Chinbok
23rd May 2007, 08:15 PM
Wes,

I used average time of winners over a large numer of races.

I think if you are going to use time ratings you have to do them properly. i.e you need track variants and ideally you would also apply a daily track variant to account for track condition, rail position and wind (I'm not doing this - too hard)

I used track variants and there can be more than 1 sec variation over 1200m from one track to another which makes a big difference.

wesmip1
24th May 2007, 08:11 AM
Chinbok,

I agree there can be huge differences and I have noted several tracks that have unusual times.

I'll get the track variants in there soon but I still have a lot of work to do.

Thanks for your help.

crash
24th May 2007, 09:54 AM
Wesmip1,

At one point I spent about 8yrs specializing in time [got some great long-shot winners too]. The horse racing mag [can't name it here but I'm sure you know the one] had a whole page of all the different aust. tracks and their time variations for distance. I'm sure if you contacted them they would be able to sell you a back issue.

Using time is a very hard way to go. A hell of a lot of work involved and like any method it has it's brickbats. Track condt. on the day, wind direction and strength, the general contd. of the track at different times of the year and even the length of the grass all affect times and then you have to adjust individual horses weight carried at various distances etc. and depending on the size of the horse [hard one to know], different weight +/- will effect their times more so or less so depending on the distance.
I had a formula method I'd worked out I could apply to each horse in any race but blowed if I can find it or remember it, as it was about 18yrs. ago I last used it.

The whole business of track times regardless of how you fine tune it, is always going to be a bit 'hows-your -father'. Still, it definitely has some great pluses going for it as a method of handicapping. It served me OK [until I tired of the work involved], especially over sprint distances. Beyond 1400m it's more about race pace and good sectionals than overall times.

Hope this helps.

wesmip1
24th May 2007, 11:53 AM
Crash,

I agree with everything you say. But I thought there would be more information out then then there currently is.

Time analysis has a lot of variables and I think most races you will be off for the various factors you have listed. I am looking at it for a place system mainly trying to back the longer priced horses that I think have the ability to run a time that "could" put them in contention.

I appreciate any help. I'll try and contact the mag and see if they know which issue it was in.

As far as work involved I am hoping to automate everything so I just have to press the button to get my time analysis for each race.

Thanks

Chrome Prince
24th May 2007, 02:11 PM
Due to the volume of work involved (a massive undertaking), I would suggest you specialise in one distance range and work your way through it until you get it near to your requirements.

The problem with this, is the number of variables, so there is not enough data to make it reliable, if it's done thoroughly.
The other problem is that they change track surfaces, rail position etc etc, so you are sometimes working blind. Even barriers are sometimes changed a few metres because of the tractor getting bogged etc etc.

By the time you hone it down to

TRACK -> DISTANCE -> GOING -> CLASS

You are left with data that is unreliable - there just aren't enough meetings with the same conditions and class to make it reliable.

The old adage used to be true "based on a minimum of 1000 races"

I can tell you from first hand experince, you need a minimum of 10,000 races.

Good luck, I'm sure you'll find a successful angle with it though.

wesmip1
24th May 2007, 04:34 PM
CP,

I have found it isn't too hard to come up with an approx rating for most of the races that is quite accurate. But it is as you and crash put it almost impossible to get a reliable figure on every horse in every race. There is usually one of two horses per race who's recent times have been affected dramatically by rail positions and the like.

But overall i think getting the order out and then elimating horses on some sound principles ( recent placings, class, jockey, etc ) you can come up with a short list which will be useful.

I have found that the order is usually around approx 70%-80% correct and it is really only 1 or 2 horses that jump more than 2 or 3 postions either way in the order.

Thanks

crash
25th May 2007, 04:57 AM
wesmip1

Over the period I used times, reasonable accuracy was surprisingly good considering the variables possible. Variables tended to even out across many bettable races and their fields of runners to a surprising degree.
In my experience, I'd say over 1000m to 1400m your more likely to be about 80 to 90% in the right ball-park and 75 to 85% for 1500m to 1600m if you get the formula of time, recent placings, class, jockey, etc. working together well. A bit of micro-adjustment here and there over a few months can tune the method up quite well.

The method does allows for a lot of winners that are definitely hookable and you would be surprised at the amount of long-shots the method throws up to the surface. Great for an e/w dabble. I would keep the method away from heavy tracks and slow tracks can be a bit dodgy but often worthwhile.

Good luck with it.

wesmip1
25th May 2007, 08:29 PM
Crash and CP,


Appreciate all the comments. I will keep at it and see what I can come up with. I still need to do a lot of work but I feel it is going to be worthwhile. I am actually learning a lot from it that will be useful for my handicapping process as well. For example it actually gives you a good understanding of the impact of the track condition on prior (and future) performances.

Thanks again and all help is much appreciated.

Good Luck.

Raven
26th May 2007, 10:11 AM
chinbok,

I have not got track variants to its full potential yet. I have noticed a few places produce very quick (or slow) times though and I do a slight adjustment for it but I would not say it is an accurate adjustment.

There are several ways to work out the track variantions and I am still condering the best way to do it. The options I have include:

1. Avg time of winner over a large number of races (500+ races).
2. Avg time of a consistent horse that has run across multiple tracks (this is good but hard to do ).
3. Track records for each distance.

Can I ask which one you prefer to use ? Depending on which one you use you are always going to get those slight differences which could cause different selections. In some races 0.1 secs can be a huge difference to whether a horse is rated as top selection or 5th selection.

Good Luck.
re Track Variants

I used to convert the time to a rating, then compare this to the "class par" rating for each race. Then simply calc the raceday ave & hey presto you have a track variant for the day. Its not rocket science, but neither is any form of ratings. This game is 100% opinion & guestimates.

Like Crash, i too gave up because of the time spent on compiling the ratings. You also have to put up with an ongoing series of problems, such as:

* No sectionals on certain days or races.
* The odd hand-timed run or runs per meeting (next to useless)
* The small amount of races per meeting to estimate the track variant
* Rain during a meeting. Times slow down, track deteriorates. On these days time comparisons are meaningless. What was the track variant on race 1? Race 8?

I have some par time info for Belmont & Ascot if interested

wesmip1
26th May 2007, 10:22 AM
Raven,

any info you can provide is appreciated.

Thanks

Michal
27th May 2007, 02:17 PM
Wesnip,

Try doing a search for horsetorque web site. The guy that runs it is Gavin H. You can purchase a book that has the
par times for most of the major track and a method how to implement them.

Good luck I think that its worth it.

Michal

PunterPete
30th June 2007, 02:33 PM
Wesnip,

Sorry I don't get on this forum much but I did see your post re Track Differences and factors for D,S etc.

I have used winning times and Last 600 times for different distances to get track models for most Metro, Prov and Cty NSW tracks. Tracks do vary substancially both for Time and 600T over distance. I strongly suggest you compare the same / similar class races to get par times for each track distance (difficult for cty tracks). My track models are for 50m incruments for each track.

I spent about 8 years adjusting mine but I now realise I should have stopped adjusting a lot earlier. As long and you are consistent in the way you create the track models and have basis models done thats enough.

What is just as important is daily varient factors (DVF) - To do these you need to compare the long term average time for good condition and same distance vs the days time and 600t for the same distance. You cannot use fixed deductions for Dead/Slow reliably. The most thorough way to get the DVF is to review each race after it is done to get the DVF.

Once you have both Daily varience factors (these can even change for each race in one day) and your track models you can do speed ratings with a bit of confidence.

Please note after over 10 years exp of doing this its still not going to make you rich unless your game enough to put large dollars down. I do make money from my analysis but its no where near as much and my day job. (You can impress your friends with it though). It does work better when you have a consistent dry season. Slow and Dead tracks luck comes in a lot more.

Good luck with it.

baco60
30th June 2007, 03:44 PM
Have a look at this,the only way is large dollars.
I got nothing to do with this web page.

ppfa.com.au/downloads/PPFA_Forum_Staker_20070628.xls