Betting teams wired for mutuel success?

[derby1592]

I read the \"Wired\" article on the Hong Kong betting teams. Very interesting (special thanks to Mall for pointing it out) and I have a few observations.

First off, I would recommend that you take everything you read in the article with a grain of salt. This early statement in the story should set off red flags for everybody, \"...horse racing is built around a pari-mutuel system...Therefore, someone with the right research capabilities can EASILY find the public’s miscalculations and exploit them for financial gain.\"

I actually like this description of handicapping in a pari-mutuel wagering environment. All but that tiny, little, very suspect word, \"EASILY.\"

These computer teams are on to something but don\'t get fooled into thinking that the current computer programs are better than a good \"human\" handicapper. The article indicated that many of these betting teams go bust and the successful ones probably earn about a 25 percent return on their wagers. This is pretty good but top human handicappers do just as well or better.

So where do these computer teams get their edge? They get it through high volume betting thanks to the large mutuel pools in Hong Kong. Their computers can tap into the electronic odds even in the complex exotic pools and identify the various \"value\" plays. They can also quickly calculate the maximum wager that will still maintain a sufficient ROI (if you bet a lot you drive the price down on your selection). The computer can also instantly make large batch wagers at the last possible instant to ensure that the odds don\'t unexpectedly change before the race is run. (It was not clear if any of the Hong Kong groups have this ability but we know that the South Dakota group was doing this before they got shut down.)

Another advantage of the computer approach (when backed by very large financial resources) is that the selection/wagering process is entirely objective and systematic. If the program is good enough to get a 25 percent return, you can be reasonably assured that you will eventually grind out that return. On the other hand, it is very hard for a person to systematically and consistently apply even the most successful human-based process and one\'s judgment can quickly get clouded in the midst of losing streak or a string of bad luck or narrow defeats.

These computer betting teams certainly raise several interesting issues. Here are two that come to mind immediately.

Longer term, what happens when these handicapping programs do start getting really good? They probably will eventually. The technology and science is changing rapidly and computer models such as these are going to keep getting smarter and smarter? Will such programs eventually \"squeeze\" all the value out of the system and turn what is now a challenging competition into a simple game of chance in which the only way you could make money would be via luck since there would no longer be any \"value\" left on the tote board?

Nearer turn and maybe of even greater concern is the prospect of \"computer trading\" causing wild odds swings after any human bettor has had the opportunity to make a wager. Imagine that you have identified a horse that you think is a good value play at 5-1 so you put your bet down just before the windows close only to see the odds plummet in the last change thanks to the \"computer traders.\" This could also turn handicapping for the average bettor into a crapshoot since they would only be guessing as to what the actual mutuel payoffs are likely to be.

Sort of scary when you think about it but then nobody ever said this game was easy.

Finally, if you think you can just ignore all this computer stuff and assume that it will all just go away, then you should consider this quote by Sir William Preece, chief engineer of the British Post Office back in 1876.

\"The Americans have need of the telephone, but we do not. We have plenty of messenger boys.\"

Chris

[Dave_K]

What is the link to this \"Wired\" thing?  The idea of a perfectly \"efficient market\" (i.e., where every horse\'s odds accurately reflect its chances) where all the value is squeezed out is a scary thing.  If that were to happen we would be better off playing roulette or blackjack (lower takes).  Can it happen?  Some think it already is happening, without the help of computers, because all the people who used to throw around silly money at the track betting their lucky numbers and what-not are gone and what\'s left are sharpies like those who frequent this board .  There\'s like a Darwinian natural selection going on that can\'t sustain itself.  I.e., in order to survive in this game you have to be a good handicapper and bettor.  If you\'re not, you won\'t survive (and you have to take up stamp collecting or lawn care as a hobby instead).  So, the bad handicappers take their silly money with them, and the only ones that survive are good handicappers/bettors whose \"good\" betting creates a more and more efficient betting market, making it harder and harder to turn a profit for anybody even themselves.  And now we have to contend with computers too?!  Well, at least there\'s talk of legalizing slots at a track near me....

[HP]

Don\'t worry, computer models have been focused on the stock market forever already and there\'s still plenty of value and \'mistakes\' to take advantage of.

Look at the people using TG and the Sheets and then look at Monarchos winning the Derby last year at 10-1 and Ink finishing second at something like 50-1. Plenty of people looking at this stuff and coming up with wildly divergent ideas, and still lots of value. Ditto for the Breeders Cup. Of course it may be tougher on a winter New York weekday with a card of six horse fields, but that\'s another story.

As long as computers get input from humans, we should be okay.  HP

[TGJB]

I would very much like to meet someone making 25% on handle. The top guys(Ernie, Richie)used to grind it out at 2-3% before rebates. Back when I was betting seriously(a long time ago)I was considered about the best spot bettor there was, handicapping 4 tracks and averaging about 6 plays a day. In my best year I made 10% on handle, and I wasn\'t hurting my own prices(bookmakers).

[Mall]

HP: Go back if you would to our exchange of 3/11. The pt I raise re the Wired article is that conceptually, the handicapping approach of the HK teams is not all that different from what a select few among sheet players are already doing on their own. The biggest difference is that the sheet players are using figs & fig handicapping concepts instead of the 130 factors of some of the HK teams.

In the 3/11 post you confirmed that, like me, you take into consideration certain factors in the context of a specific race to make a judgment re whether each horse is an overlay to go fwd, bkwds, or remain the same. As I understand it, that\'s what the cutting edge players are doing as well, except they are not content to do a rough calculation in their head.

The programs they use start with a judgment regarding each horse\'s stage of development & approx. % chance of improving, pairing, and/or running a gd no. in relation to the horse\'s effective top. This info, along with the effective top itself, the horse\'s weight(don\'t even think of going there), & the horse\'s post position are used in a large no. simulation to determine how many times each horse would win if the race was run, say, 20k times, which is then used to calc a true odds line.

A major advantage of this approach is that it forces you to make decisions which should be made in evey race. It also allows you to take advantage of a database which even the most experienced player cannot match, while getting the benefit of the \"law of large nos.\" Pretty damn clever, if you ask me.

[Dave_K]

Mall, you raise a very interesting issue in your last post, and especially with the line, \"The biggest difference is that the sheet players are using figs & fig handicapping concepts instead of the 130 factors of some of the HK teams.\"  The issue is one also raised by Mark Cramer (in my opinion, one of the most brilliant handicapping authors out there) in one of his books (I forget which one): is it better to focus on a few (or even a single) factor that has what Cramer calls \"wager value\" (e.g., TG fig patters) or factor every piece of information relevant (maybe even more than 130 factors) to the race outcome into the handicapping decision?  To paraphrase Cramer, \"If you find a profitable angle, don\'t mess up a good thing with \'handicapping\'.  As the number of factors that go into a handicapping decision increase, the more that decision will converge with the betting public\'s decision\".  Interesting ideas (I think).

[HP]

Mall: I guess what you are saying is that I am a cutting edge handicapper, or at least as much as I can be considering less than 130 factors. The only other thing I can think of to say about this is that I would have no interest in being involved in one of these \'team\' things.

Kearns: I agree with Cramer that one good angle can be worth as much as 130 factors. There may be even MORE factors to consider, but I think the point is to figure out which ones are more/most important in a given situation.

I knew one guy in my old neighborhood who had an interesting idea-he would pick one type of race (MSW dirt routes, Alw dirt sprints) and just keep looking at them throughout a meet, and he was a winner. This kind of \'narrowing\' approach (whether it\'s the type of race or the number of factors considered) interests me more than loading up (or loading up a computer) on everything I can think of to bet a race.

I\'ve had more success having an idea and going with it; many times I\'ve had a good initial thought and then as the race gets closer and I keep looking at stuff and weighing factors I end up someplace else and losing. HP

[Dave_K]

I\'ve finally read the \"Wired\" article and would like to make two observations.  First, I remember looking at Benter\'s paper in \"Efficiency of Racetrack Betting Markets\", where he describes what he\'s doing in Hong Kong and wasn\'t all that impressed.  The kinds of data he was putting into his formulas were pretty simplistic, like \"Jockey\'s win %\", \"Lifetime earnings\", etc, etc.  Secondly, didn\'t Bill Quirin do essentially the same thing as these guys are doing with his regression equations like 25 years ago?