Question for Software Makers

[DeanT]

I have demo'd a couple of mechanical systems, and who isn't looking for an easy way to make contenders with little hard handicapping. One of the drawbacks I see in mechanical selection in North American racing is trainer changes, bad lines and trainer intent in specific races. There are so many question marks. We can hope to smooth them, but they are a drawback, imo.

Benter and Woods et al in Hong Kong had powerful programs and Hong Kong had several things going for it: Bad lines were explained with data (the reason for the bad lines are published); trainer changes were very minimal, the horse population was small, and when horses are entered they run true to form. No Joe Blow to Scott Lake form reversals.

I am wondering, since I am a fan of much of the software out there: How good could yours be with a more formful horse population, with 7 day a week detention barns where you can be sure there is no funny stuff, no trainer changes and the like? Could your software rival what was done in Hong Kong? Would your hit rates for your ratings (speed or pace) go up by a significant amount? Would your "bad" error be almost eliminated? How much better would your program be in such an environment?

[shanta]

Dean
I believe Ken Massa over at Htr is already working on readying his software to operate in the Hong Kong market.

I really like your "picture" of how things are over there man.
Richie

[Sinner369]

I agree.......with today's technology.....Surprised that the software guys haven't come up even with a mini version of their software built in the 80's and early 90's!

We seem to be copying his (Benter's) theme.............Create own True Odds line.....look for overlays by comparing true odds to public odds.

They are making millions............why can't we create something that make.........thousands???????????


sinner ;)

[DeanT]

Dean
I believe Ken Massa over at Htr is already working on readying his software to operate in the Hong Kong market.

I really like your "picture" of how things are over there man.
Richie
Thanks Richie. I figured with it coming in now, someone would jump on it.

Have you ever really sat there and played a card? It is a thing of beauty, imo. If you bet physicalilty, think again - every horse is completely sound, and every coat is dappled. You simply can not get into the box if you are not sound and fit to race. Eghad. Joy for a horseplayer who simply wants to bet a nicely crafted pari-mutuel puzzle.

[shanta]

Thanks Richie. I figured with it coming in now, someone would jump on it.

Have you ever really sat there and played a card? It is a thing of beauty, imo. If you bet physicalilty, think again - every horse is completely sound, and every coat is dappled. You simply can not get into the box if you are not sound and fit to race. Eghad. Joy for a horseplayer who simply wants to bet a nicely crafted pari-mutuel puzzle.

Dean
To be honest I have never seen a race from there man. You have me all juiced up to see some now for sure! :ThmbUp:

[DeanT]

I just dabbled with it Richie. It is on TV here late at night, so I throw it on and mess around. Small pools because there is no commingling so I mainly focused on potential arb opportunities. They get some patrons. Chinatown's in Toronto and Vancouver are pretty big. There is an OTB in downtown Toronto which is pretty packed late at night in C-town here.

I did some research on it and found a few websites who have their proprietary speed and pace ratings. Before they took it off betfair I used to watch to see what some of the sharpies were doing as well.

When it is here I sincerely hope we can get data files. I would love to research it. Which is part of the reason for starting the topic.

Some highlights I found over the past couple of years:

* The vet list is published on the HKJC website, so if a horse comes in off time off, the reason can be found - scoped with mucous, lame in front, etc. Good way to find out why a horse might have been off form, or missed time. I honestly think the Benter programs might be a bit inefficient with this type stuff, but maybe not.

* The horses are all under lockdown. Vets can get in and out, but any medications have to be logged in a book. Anyone who messes with that are banned. No milkshaking that's for sure. 24 hour video surveillance.

* The British feed is top notch. Time between races is pretty big. They show replays of the contenders and concentrate on them in the paddock/ring.

* If you've been drinking, the Hong Kong feed is fun to listen to :)

* Field size is big. Races are contentious.

* Every horse is as beautiful and sound as a horse can look. Don't even bother looking for lameness.

Anyhow, that's my take. I might watch it tonight for awhile. Japan comes on at 1AM as well.

[shanta]

Dean
Where can you watch the races? I have Brisbet and they are not listed there man.

Are they running a meet now?
Thanks

[DeanT]

Sha Tin (HK) and Kyoto (JPN) are running now and are bettable and watchable here in Canada. No comminlging, tho. I heard that commingling will be coming soon.

I think they have some sample races here, but I am not sure if you can watch live video.

http://www.hkjc.com/english/hrc/hindex.asp

Post time is 11:30ET and 1AM ET for HK and Japan respectively (this evening). I thought someone in Vegas was bringing HK in to the US market?

[shanta]

Sha Tin (HK) and Kyoto (JPN) are running now and are bettable and watchable here in Canada. No comminlging, tho. I heard that commingling will be coming soon.

I think they have some sample races here, but I am not sure if you can watch live video.

http://www.hkjc.com/english/hrc/hindex.asp

Post time is 11:30ET and 1AM ET for HK and Japan respectively (this evening). I thought someone in Vegas was bringing HK in to the US market?

Thanks Dean for the link. let me check it out. I think the Vegas connection thing kicks in a couple of months down the pipe but not totally sure. I'll check it out.

Thank you again for taking the time to help
Richie

[46zilzal]

We carry both signals and it is amazing to see the crowd change in the race book after 8:30 PM when they start. One section of the race book keeps the Chinese commentary from that satellite and it is jammed with people who speak the language and they bet BIG. I walked by a table and saw over $5k of tickets loose there.

[Jeff P]

Dean, Great Question!!!

You can model trainer changes (XTHSTARTFORTRAINER) and bad lines (FORMCYCLE) into your UPR odds line with JCapper. You can also model trainer intent by creating trainer specific UDMs to take advantage of the strengths and weaknesses of individual trainers. The functionality is certainly there if you want to pursue it.

Keep in mind that Benter & Co were/are essentially living eating and breathing horses and following and modeling one circuit instead of dozens (or hundreds.) That one circuit has huge pools, purses, and fields. Contention runs deep in every race. Lots of fan interest (and money) too. Kind of like Breeder's Cup Day here... but almost every day of the week there. The money they make/made is proportionate to pool size... how much they can bet without distorting their own payoffs. If you are successful in North America it must be on a smaller scale because of the smaller pool sizes we have here. But your own degree of success CAN BE proportionate.

IMHO, the following one circuit part is a huge difference maker. Every racing surface has its own tendencies from an energy distribution and human connections standpoint. Here in North America us software makers take on the role of producing (or attempting to produce) algorithms and ratings that perform well across all racing surfaces. The fact that some of the ratings I've seen (and produced) perform as well as they do across all racing surfaces is no small feat.

But performing reasonably well across all racing surfaces is something Benter & Co never had to consider. They are/were able to tailor their algorithms for a tiny handful of racing surfaces. And THAT is something far easier to accomplish than the one size fits all activity that us software makers are judged on.

I'll make the argument that the player who truly takes the time to do the requisite work, and through research becomes extremely knowledgeable about the intricacies of any one racing circuit... the energy requirements (early/late) of the surface... positional rail biases from gate placement and course layout... habits and tendencies of the trainer and rider colony... he or she will very likely find they can model their own specific knowledge into UDMs and UPR tailored to that one circuit and achieve results at the windows that would be surprising to most.

One of the downfalls most players face - and I am as guilty of this as much as anyone else - is that we tend to spread ourselves too thin.

To answer your question, if North America had only one racing surface, I would have very little trouble producing an auto generated UPR where the top ranked horse in each race was very close to flat bet break even in the win pool. And from there, very little trouble producing UDMs that could be bet profitably in a mechanical manner against an auto generated UPR odds line.


-jp

.

[DeanT]

Thanks Jeff. I am a big proponent of your trainer based modelling. It is the bulk of my work for sure. I love it.

Now, hypothetical........ can this be done.

Specific trainer moves, and noise can wildly change an odds line, correct? For example, Scott Lake to a trainer with a 0% win percentage who ran a 90 Bris figure might get a rating of 70 in your JPR and odds line of 3-1, or might get a similar rating in HSH or whatever, based on his ability in his last five lines. But that horse, it can be safe to assume, might run 10 points less this time because of the trainer move. Similarly, a horse coming 2nd time off a layoff with Dickinson versus Pletcher are two completely different animals as well. UDM's can point to these as plays, and add noise to a computer generated odds line focused on ability only.

My question: If I give you an inventory of only 30 trainers, and 1200 horses like Hong Kong, can you build specific trainer patterns, and other noise directly into a JPR number, and subsequent odds line? Meaning (perhaps more clearly) you run the numbers and get an odds line. Horse A's rating is a 60, and he is off a 45 day layoff, where this trainers horses run new fig tops several times over the last five years off that, so your rating after putting it through this filter is now a 70 rather than a 60. In the exact same race there could be a horse who has a trainer who does not have the horse fit off the move, so his 60 might get downgraded to a 50. The ten point bump is solely based on a trainer specific UDM, which gets quantified.

Is this possible in a small inventory of horses and trainers like Hong Kong? Can this stuff be built right into a JPR number?

[Jeff P]

Dean, to answer your question: YES... something like this could be done in a custom (or future) version of JCapper. What it would require from a user interface standpoint is a checkbox in the UDM Wizard to toggle/enable this feature on an individual UDM basis along with a separate field where the user keys in an impact value for the UDM. Then during number crunching the UDM impact value is used to adjust the power rating and subsequent odds line.

-jp

.

[DeanT]

I would be interested in that Jeff.

That is one thing that is neat in NA t-bred racing - a race at LRL and DEL can be two completely different animals, based on trainer colony, track bias etc. We seem to be so amazed at the computer groups overseas - like they are some sort of messiah - but maybe it is unfounded. Your program, and Dave S's and others might be equally as powerful without all the noise we encounter here. I would love to see builders like yourself and Dave have a shot to write something for the HKJC. It would be interesting. I would love to be able to see your JPR w% right now if you can ever get historical data. I think your assumption (and mine) that your hit rate would be better might be true.

[equicom]

Hi Dean,

I'll probably post up a more detailed reply later to your original question, but just wanted to mention for now that if you're considering the Asian racing situation, Japanese racing IMHO is much more reliable and honest than Hong Kong racing. Usually Japanese horses are always out there to win, and the trainers are fairly competent, the jockeys are slowly getting there too. In Japan, all racing is controlled by the govt and trainers are actually in a sense "public servants" and government employees (probably amongst the best paid in the world, however).

The situation in Hong Kong is very different. Many of the better jockeys are imported and riding on permits, and more than a few of them are as crooked as a crocodile with three legs. The trainers are usually toffee-nosed aristocrats with names like "Sir Basil Cleverclogs" and they're often even more crooked. The aristocratic types are often the worst of humanity while portraying themselves as the best. It's one of the reasons why you guys had the American Revolution, to get rid of those bstrds!

Anyway, in HK, there is huge amounts of money flying around and all the corruption that goes with it. Sure the jockeys usually ride to instruction, but the problem is you never know whose instructions!

Brad

[Dave Schwartz]

I'll probably post up a more detailed reply later to your original question, but just wanted to mention for now that if you're considering the Asian racing situation, Japanese racing IMHO is much more reliable and honest than Hong Kong racing.

I would concur with this statement.

Besides - bigger, more stable handle.


Equicom,

Has anyone conquered the data files? Last time I worked on the Japanese files they were like 14,000 fields spread across 12 daily files and the instructions were all in Japanese.


Regards,
Dave Schwartz

[DeanT]

Very interesting Brad. I had never heard that before.

Do you play either place? Can you model the "scum" like we can at places here, or are you up against it?

As I had mentioned earlier, I can't seem to bet physicality there, because all horses look amazing. I can't use data files. I cant find any trainer patterns in my cursory look. Plenty of swimming sharks, too. No betfair. So, nope, I think I will pass. After reading what you posted, I might be smart to pass.

[equicom]

Hi fellas,

I can't speak for anyone else because I don't know at this stage what else is out there. What I can tell you is that as far as my own service is concerned, we'll be rolling out Japan data in the medium to long-term future. Whether those data files would be of any use with Dave's product or work with the SCUM analysis model is not known at this point in time.

Dave, if you're listening, I'd certainly be interested in discussing at some stage with you what we can do to make the data compatible with your software. I do have to say that the structure of our data files is considerably different to the structure of other providers, so there's a high probability that it would not work in the raw form, it would need some sort of conversion to work with your software. We'll actually have people on the ground in Japan putting this together for us, and it will be in English!

I do actually have firsthand experience in the industry over there, of course a long time ago and they've really come on a long way since then, but the key to understanding Japanese racing form has a lot to do with the tracks. American racing tracks do have their differences, but the variation in shape and circumference of Japanese tracks is much greater than in the US.

They also work their horses much harder than we do and in all kinds of weather and on all surfaces, including snow and I've even seen them gallop on rubber and work on bitumen. They are incredibly tough horses and there's a lot of good blood there, mostly from Sunday Silence, but you also get Woodman, Halo, Storm Cat, etc.

Incidentally, if you want to see what a difference it makes having vast amounts of government money to throw around rather than having to rely on private funding, check this out (you'll need hi-speed connection!)...

http://www.youtube.com/watch?v=4SPUkBZOeWk&feature=related

If that doesn't make you green with envy, I don't know what will.

Brad :cool: (green with envy, as you can see, but trying to be cool about it!)

[098poi]

I would concur with this statement.

Besides - bigger, more stable handle.


Equicom,

Has anyone conquered the data files? Last time I worked on the Japanese files they were like 14,000 fields spread across 12 daily files and the instructions were all in Japanese.


Regards,
Dave Schwartz

I'm surprised you would let a little thing like "instructions in Japanese" stop you! :D

[njcurveball]

I'm surprised you would let a little thing like "instructions in Japanese" stop you! :D

From what I know of Dave, this is a pretty good joke! :ThmbUp:

I cannot think of a harder working man in any industry! :ThmbUp: