OZmium Sports Betting and Horse Racing Forums

cheekyshiraz

31st March 2010, 09:59 AM

darkydog2002

31st March 2010, 11:32 AM

Why use Time ratings at all.
If the best time rating bloke in the world - Andy Beyer) couldn,t make them work on Australian tracks and the Official race club handicapper disregards them in favor of class and weight it seems to me a total waste of your time.

Cheers.
darky

Pauls123

31st March 2010, 08:17 PM

Hi again Cheekyshiraz,

For what its worth, I use the basis of 1 length = 1 1/2 kgs. And there is 6 lengths to one second. So in your first example horse B would beat horse A by .6 of a length. So I calculate horse A would run that 1000m in around 59.1

As to your second example,....not sure of formulas to differentiate good from heavy etc. My formulas are very general.

Good luck, Paul

Chrome Prince

2nd April 2010, 11:31 AM

Hi guys

Im just trying to do some calculations to see if i can more accurately predict the winners....i know horse racing is not a science...its more difficult than that :) however, if you can shed any light on the following questions i would be obliged.

How many lengths on average is a horse handicapped by per extra kilo? so say 2 horses were carrying 55kg for a 1000m and both ran it in .59s. if we put and extra kilo on horse A before the start...what would we expect his time to be then or how many lenghts behind horse B?

How much slower on average would horse carrying the same weight run a 1000 in different conditions. So say the average time for 1000m on a dead 4 track for horse X is 59.5. What time could we expect on a good 4 track, a slow 4 and a heavy 4?
I understand that some horses prefer different conditions but is it right to assume that most horses will run faster on a dead track and slower on the heavier?

would be great to hear your opinions and analysis on this.

cheers

Can't shed too much light on this, but I'm working (slowly) on a time ratings method to incorporate into my database. I'm extremely dissatisfied with the way in which time ratings are worked out by most places. I'm taking a different approach to calculations.

For example par times don't handicap (either way) enough.
Beyer failed because our pace and surface is different to the USA.
If he had bothered to put in some effort, he'd have realised adjustments need to be made, but the principles are sound.

USA races tend to be run at a faster clip than here, they tend to go all out and it's survival of the fittest and fastest, whereas here it tends to be a calculated pace and the horse with the fastest final burst often wins or goes close.
In the UK it tends to be even worse, the pace is even slower and final bursts are faster.
It's quite a culture shock.

Time "ratings" is a science, a fairly precise science, which is why a lot of people get it wrong. I'm not saying I'll get it right either, as missing information kills a lot of what I had intended, but I think I'm closer than I thought I'd be.

To answer the latter part of your post, it will all be different.
Times vary track to track, distance to distance and going to going.
Each is a subset of the other.
I've found it's dangerous to take a blanket approach.
Good to Dead 1000m at one track is not the same as Good to Dead at another.
Apart from the best time ever recorded under those conditions, the slowing down effect will also vary.
It might be 1 second at one track, and 2 seconds at another.
Then there's rail position which can impact the time dramatically and pace.

A lot to chew over, but what started out as fun has become an arduos task.

Pauls123

2nd April 2010, 12:28 PM

Thanks for that insight Chrome. I always find your posts fascinating and interesting reading. Looking forward to whenever you come up with some time ratings strategy in your database.

I still find the only way I can win money over a period of time is to mess with the sectional times. I used to do melbourne and sydney, gave melbourne away to just concentrate on sydney only, and what happens,.......now you cant get rosehill anymore, what a pain.

Paul

lomaca

2nd April 2010, 12:45 PM

Hi guys
How much slower on average would horse carrying the same weight run a 1000 in different conditions. So say the average time for 1000m on a dead 4 track for horse X is 59.5. What time could we expect on a good 4 track, a slow 4 and a heavy 4? According to Paul Segar a general approximation of different track condition and distances:

1200 M Good 1.11 Dead 1.13 Slow 1.15 Heavy 1.17

1400 M Good 1.24 Dead 1.26 Slow 1.28 Heavy 1.30

1600 M Good 1.37 Dead 1.39 Slow 1.41 Heavy 1.45

2000 M Good 2.03 Dead 2.05 Slow 2.08 Heavy 2.12

(that's minutes)

Chrome Prince

2nd April 2010, 03:01 PM

It's not really a plug for my database, because I don't know how it will actually turn out, whether it will work out or not.
But I still believe that using specifically tailored data is better than benchmark data.

As an example, par times do not really account for pace.
Par times tell you averages, but don't tell you how good a run was.
Using track records over every distance in every track condition is more useful in showing class, and how far off that time a horse ran.
I am a firm believer in sectionals, although race sectionals are not anywhere near as good as horse sectionals.

A poster posted this in another forum to do with time ratings:

The track record is 86.5 (set long time ago) I have allowed 2-3 lengths outside that (split the difference between the run by Regansburg at this meeting 2 yrs ago and the track record) By using quick times we are identifying horses that can actually run time and get the rating they deserve rather than be "soft" and allow all the crabs to be over valued.It is this principal that has been applied, a horse doesnt know what class is is, only that it can run, how fast is the key and "fast" should be rewarded and mediocre should not.

With all due respect to the poster and the product, allowing 2 to 3 lengths off a track record is in my opinion doomed. 2 to 3 lengths is the difference between an average horse and a great horse, I don't see making allowances as doing the ratings, the horses nor the punters any favours.
A fast race time is not the only benchmark to class either, it must be a portion of an overall picture.

Some of the fastest race times over distance are won by the most ordinary gallopers, it's the time run after a certain pace that tells the story.

As I say, it's a bit of a riddle, but I think you must start with exact parameters, for there to be any chance of success. One second difference is too major a factor to ignore in my opinion.

I'll make some illustrations to demonstrate the point when my database is sewn back together.

cheekyshiraz

2nd April 2010, 07:31 PM

Chrome Prince

2nd April 2010, 07:46 PM

Yes, that product has par times inbuilt now.
Not sure how accurate they are, as I actually have that software myself and have only seen a small improvement over a significant number of races.
I'm currently working on a different approach myself to times, but not sure where it will lead.
I do know that their par times are approximate estimates and this correlates to the results.

lomaca

2nd April 2010, 08:02 PM

Thanks guys for your thoughts and info. ya i ve seen as chrome pointed out that the par times differ also from trace to track so a 1200 par time will differ from rosehill to caulfield to moe. So i guess its best to try and concentrate a method on one or two tracks and see how the results turn out.

is there any handicapping software on the market that already has all the par times per track and allows the user to input variables such as weight, barrier No. etc? Ive heard of bet selector but have nt really looked into it.

cheerscheekyshiraz

On the face of it time rating should be the bee's knees in racing,
after all the fastest horse wins the race.

I have persevered for years with time rating. Have my own par time tables for every race track in OZ, worked out of actual race times ran on the track, for every distance under all conditions.

I must have been doing something wrong because I never could make time rating pay. I sure picked up a lot of good prized winners but not enough for a profit.

My combined class, weight and other ratings on the other hand, kept me in style for years now.

Betselector is basically time rating.
Tried it for a while early on but I was doing far better without it.
I don't know how it performs lately, maybe improved?

If you want my par times give me an email addy and I shoot them off to you.

Chrome Prince

3rd April 2010, 09:37 AM

I think everyone struggles to make times pay at one point or another.
To be candid, the other product has a newer way of calculating pars which has seen somewhere in the vicinity of a 4% to 6% improvement.

I'm not convinced they are doing it the right way though.
A tell tale is in the odds generated for their pp market.

I don't want to go into this much more, as I'm not a product knocker.
I personally have it and it suits a purpose, but in my opinion there are a number of shortcomings I'll keep to myself.

Onto other things.

The trots are a very good learning place about the impact time.

Two horses can run a mile in 1:59.90, one can be a champion, the other a battler from the bush. The key is in knowing how to rate the same exact times differently.
If one used 1:59.90, you'd have them rated equally, one will be well unders, the other well overs.
One will continue to win races, the other may win one more race in it's career.
Given that the trots are run a little differently, the main principles are the same.
Running time tells you the race was a competitive race, but it doesn't tell you if the horse won running away or won by a lip under the persuader.
It doesn't account for position in running, if it swooped from the back or was taken on up front, or left alone up front and dictated the whole race.

Of the two 1:59.90 horses, it is crucial to know how fast they did their early splits in and how fast they came home.

They could have jogged early and flew home for a sizzling final quarter adding to an overalll quick time.

OR they could have gone hard early, took a breather mid race and sizzled home.

OR they could have sizzled early and plodded home

All having the same race time.

Pauls123

4th April 2010, 07:01 AM

lomaca

4th April 2010, 07:17 AM

Hi there Iomaca,

I'd love to have a look at these par times of yours. Any chance I could be included in that email please.

Much appreciated.

Paul...............................email addy
pauls @ tsn.ccis that just a com or is there an au at the end as well.

I use Oracle database, so I can only send it as an Excel or text file.

cheekyshiraz

4th April 2010, 10:20 AM

If you want my par times give me an email addy and I shoot them off to you.Hi Iomaca

I would love to see the par times you have. I have tried looking for your email but cannot access it. mine is adie @ emeraldideas . net

I will try a few different methods and systems with them to see what i can come up with.

it would be fantastic if there was a software that had a database of all horse racing results over the last 10 year in australia with all the data (barrier no, weight, jockey,track, etc) for each horse in the race with All the times the races were run in. That way if you come up with a new system you can test it on the old races to see what selections would have been chosen and you could see what S/R would have been achieved.

cheers

cheekyshiraz

4th April 2010, 10:29 AM

Of the two 1:59.90 horses, it is crucial to know how fast they did their early splits in and how fast they came home.

They could have jogged early and flew home for a sizzling final quarter adding to an overalll quick time.

OR they could have gone hard early, took a breather mid race and sizzled home.

OR they could have sizzled early and plodded home

All having the same race time. [QUOTE]Hi Chrome

 The pace of the race is critical as you illustrate with the trots but i would like to just concentrate on the sprits of up to 1200. This way you cut out a lot more research as would be needed for the longer races as there would be many more sectional to study and pace is a bigger factor. (no breather mid race etc) I think par times for the shorter distance reflect a more accurate time than those run over the longer distance therefore giving a more accurate selection.

I may be wrong with this assumption as i do not have a database to test this but this is what i will be looking into with the par times.

cheers

lomaca

4th April 2010, 10:56 AM

Hi Iomaca

I would love to see the par times you have. I have tried looking for your email but cannot access it. mine is adie @ emeraldideas . net

I will try a few different methods and systems with them to see what i can come up with.

it would be fantastic if there was a software that had a database of all horse racing results over the last 10 year in australia with all the data (barrier no, weight, jockey,track, etc) for each horse in the race with All the times the races were run in. That way if you come up with a new system you can test it on the old races to see what selections would have been chosen and you could see what S/R would have been achieved.

cheersI have such a software but it's not for sale.
You can buy BetSelector, reasonably priced or a much more expensive one, can't recall the name, supposed to be the best.

ChromeP can tell you about the former and I think Bahgwan has the other one.

Files sent, 5 Excel files not too big, should go through.

Chrome Prince

4th April 2010, 05:27 PM

Let me be perfectly candid here and reveal all my cards, as I think it's only fair with what I'm about to post.

I own a copy of Bet Selector Platinum Professional
I developed and market RaceCensus Horseracing Database with the help of this site.
Both products have pluses and minuses to them.

One product will suit some type of punters and the other is an alternative to the other.
Both products are tools to help punters, and both have their place in the market.

I'll try and explain the main differences:

RaceCensus was originally developed as a cheaper alternative and a bare bones system tester that had past data and didn't cost the earth.

The other product takes an extremely long time to analyse results, because ratings are calculated on the fly. So to change one parameter and rerun a test takes in excess of 9 solid hours, even after a computer defrag.
However, the other product can download form and results and automatically parse the data.

Since time has gone on and RaceCensus has developed a little and so has PPP, I became a little unsatisfied with the way in which ratings were calculated. The forecast prices became extraordinarily unrealistic at times.
However, PPP has it's place and is a good product, if you want print outs of the days bets etc.

RaceCensus by comparison cannot download form and results, but can be updated monthly via importing a simple spreadsheet.
By comparison, system tests are lightning fast, for example a test in one program takes 9 hours and 45 minutes on a Pentium 3Ghz with 2GB ram.
Racecensus takes 2 minutes 45 seconds (once the records have been indexed for the first time). This is a massive difference when making adjustments to system tests along the way and re running them.

RaceCensus is currently being updated with Metro AND Country races since 1998. Many errors are picked up by my program that go undetected in others, because I am alerted to index mismatches or missing data.

Nothing is added to the program that hasn't been either popular request, or is beneficial to the punter (i.e. it works).

Example: I'll never price an unraced horse at 1000/1 not knowing anything about it at all, it seems totally illogical to me.

But different people have different expectations and requirements.

I am currently working on "time" as I've mentioned, but won't add it until I'm perfectly satisfied with it's performance over years of data.

Pauls123

4th April 2010, 05:40 PM

Hi Iomaca,

Thanks again for your reply. My email addy is pauls@tsn.cc

No coms or au's or anything like that. Just,............

pauls @ tsn.cc

Thanks again, and excel is fine.

Paul

lomaca

4th April 2010, 05:50 PM

Hi Iomaca,

Thanks again for your reply. My email addy is pauls@tsn.cc

No coms or au's or anything like that. Just,............

pauls @ tsn.cc

Thanks again, and excel is fine.

PaulOn its way.

Good luck

cheekyshiraz

4th April 2010, 07:23 PM

I have such a software but it's not for sale.
You can buy BetSelector, reasonably priced or a much more expensive one, can't recall the name, supposed to be the best.

ChromeP can tell you about the former and I think Bahgwan has the other one.

Files sent, 5 Excel files not too big, should go through.Thanks for that Iomaca. I would like a software that would have or allow you to imput a handicap of lenghts converted to time for each variable of the race and add it on to the estimated time for a particualar horse...for example:

Say we looking at a race at Rosehill over 1200m on a Good track. Horse No. 1 has 1 previous start over this distance and ran 1.5L 2nd in that race from barrier 1 and the race was run in .59.5 - We want the software to calculate the est time he ran that race as we only have the time of the winner. So the program computes his time as 59.25. (.25 for 1.5L..we should be able to enter our own estimate also and the length time must differ for each track condition 7L on heavy will be slower that 7L on good etc)

We then want the programme to add on an estimated additional time for the new race. So we enter the current details of the horse for this race...he is carrying 2kg more and starting from barrier 10 racing on dead track. so we want the software to add an extra 1.5 seconds to his est time for the dead track condition and say .16s (1 length) for his wider barrier position and .30 for the extra weight. The programme would then give us an est time (in this example 1.01.21) for the current race for each horse based on there past and current details.

Is there any software out there that has the capability to do this?

cheers

cheekyshiraz

4th April 2010, 07:31 PM

I am currently working on "time" as I've mentioned, but won't add it until I'm perfectly satisfied with it's performance over years of data.Hi chrome prince

Just read this post after my last post below...can you advise if there is such a programme that has the capabilities that i outlined in that post?

cheers

Chrome Prince

4th April 2010, 08:37 PM

I'm working on pretty much exactly what you outlined except for the weight carried.

I have over 10 million horse runs in a database, and am currently looking at rating horses as follows...
Take the track record over each distance and each going.
Take the actual race time of the last race.
Convert the winner's time to a rating off the same record.
For each horse, by beaten lengths, calculate the time off track record.

In ths way, I can line up a 1200m Dead race at Rosehill with a 1400m Good race at Randwick.
A clear picture should emerge of horses rated by time ranking.
I will be using actual individual saddle transponder info to convert the beaten lengths over various distances to time.
So no standard par or set formulae, as they change depending on pace, distance and going.
Rough initial testing finds that this method of time handicapping is fine regardless of the weight carried.
I know many will throw up their hands at this, but the figures seem to suggest that weight is a fairly insignificant factor.
Perhaps over the real distance races, it has a bigger impact, but the shorter sprints seem to suggest, that weight is minor.

Michal

5th April 2010, 07:39 AM

Hi Chrome,

Where do you find individual sadle transponder times ? Isnt this something only available at a few tracks and a fairy tale for majority of the selections that you would be rating ? Personaly I wish it wasent , as we as punters and main income providers to the whole show are treated as wood ducks, feed scraps of error ridled information that was state of the art 30 years ago and hasent changed much since.

Michal

cheekyshiraz

5th April 2010, 08:39 AM

I'm working on pretty much exactly what you outlined except for the weight carried.Hi Chrome

Thats great to hear...when do you expect to have this complete? will is be part of a new version of your racecensus software? I would hear when this project if finished.

cheers

Chrome Prince

5th April 2010, 10:40 AM

Hopefully I'll be releasing version 7 by the end of the week.
The times part will be done for W.A. by the end of the month.
If there is an advantage fow W.A., I'll move on to do the rest of them.

The sectionals with transponders are available rom Sky, TVN etc.
I don't need it for all tracks, I just need it for the conversion of lengths to time.

Pluggerduck

23rd April 2010, 02:19 AM

Have you looked at this site ? Daniel has his own speed ratings and website and couldn't reccommend him more highly. His own database and ratings system was built and constructed over a long period of time. I have no connection to the site except for being a customer / his ratings have suggested good and bad betting races and has completly changed my betting losses into profits. Actually i think he used to post on here as O'SULLDJ. Very smart cookie but only does metropolitan meetings, i have emailed him and asked him about provincial races and he responded with a maybe in the future.

If the worst thing you do is join for free and read his art**********************:-)

john55

6th July 2010, 10:04 AM

Hi

I am new to this forum. I am fiddling around with time ratings and found the different points of view on this thread interesting. I am slowly putting together a set of my own par times and will probably be restricting my use of these par times to races of 1600 metres or less, as I feel time ratings for distances any greater than 1600 metres are subject to too much variation.

A problem I see though is that a race over 1600 metres could still have horses in it that have recently raced over 1800 or 2000 metres and knowing the value of these runs can be important. The problem with calculating a time rating for a distance race is that there can be great variations in the winning times (due to lack of pace, etc.) which leads to great variations in the time rating (when compared to the variations in ratings for shorter distances).

At the other end of the distance scale, I have also noticed that winning times for 1000 metre races seem to be bunched quite a bit closer together than the winning times for races over 1200 and 1400 metres, meaning ratings for 1000 metre races may end up being greater than they should be (when comparing them to ratings for 1200 and 1400 metre races).

What I am wondering is: do people that have developed and used their own time ratings make allowances in their calculations for the distance the time rating has been run over (apart from the fact that there is a small difference in the time per length as the distance increases or decreases) ?

I hope I have explained myself well enough here