Software Learning Curve

[traynor]

Originally Posted by PaceAdvantage

And here I took it to mean that the only reason there are "whales" is because they were all wealthy to begin with, primarily from inheritance...I was under the impression that Traynor believes there are no true whales making serious, serious money.

Quite the opposite.

[Dave Schwartz]

No they weren't all whales to begin with. They don't all have enormous staffs either. The one from Canada stated on 60 minutes he had a staff of twenty and he bets 300 mil a year.

I know that operation.

I will stop talking now. You guys can choose to believe whatever you want.

I will tell you that the smallest of the 6 has 45 employees. Despite the fact that he says (publicly) that he started with just him and a laptop, he actually began with a "study, analysis and programming team" of over 20.

The largest has 175 employees.

Believe what you will.

[Capper Al]

Originally Posted by raybo

Define "mini rollers".

Under $20 a race.

[Capper Al]

Originally Posted by Dave Schwartz

I know that operation.

I will stop talking now. You guys can choose to believe whatever you want.

I will tell you that the smallest of the 6 has 45 employees. Despite the fact that he says (publicly) that he started with just him and a laptop, he actually began with a "study, analysis and programming team" of over 20.

The largest has 175 employees.

Believe what you will.

Ouch!

[traynor]

I think an unnecessary layer of confusion is being created by the use of term "whale" (by almost everyone else) to refer to "large bettors" and by Dave to refer only to six specific people.

Similarly, I think it is confusing to consider that all "large bettors" (or more specifically) "all large bettors winning significant profits" are ONLY doing so with the same (at least generic) type of software, and that those "large bettors winning significant profits" are ONLY able to do so with the aid of 10-20 "helpers" and mega-millions in bankroll (apparently required by that particular type of software).

There are bettors winning significant amounts of money (with "significant amounts of money" operationalized as "profit in excess of $500k a year") using software that--with conditional wagering at an ADW--does NOT require anyone other than the user to use, does NOT require a flurry of last minute activity based on the tote board, and does NOT require more than an hour or so a day to operate. At tracks that support fairly large wagers without (substantially) diminishing mutuels, users of such software may choose to wager onsite. That may involve more time invested on any given day, but could hardly be considered grueling or unpleasant (unless one detests racetracks).

[traynor]

Do I think it is possible for a solo bettor starting with a relatively modest bankroll to win enough profit (meaning way more than a couple of percentage points shy of break even that requires rebates to be "profitable") to build up fairly quickly to "professional-level wagering"? Absolutely.

Because one specific type of software (apparently) is seriously constrained and requires (whatever) does NOT mean that ALL software has the same constraints. As Katherine Jung observes, "That is a conceptually impoverished point of view." Not to mention extremely limiting.

If someone has a "professional-level" piece (or pieces) of software that cannot generate enough profit to double a bankroll in a month or six weeks, and continue re-doubling until a level is reached that additional increases in wager size is counterproductive within (at most) a year (with six months a more realistic target), the developer of that software should keep working on it until it is finished.

[whodoyoulike]

Originally Posted by traynor

I refer you to dl's thread, in which he fully explains what he is doing and what he intends to do.

Haven't seen any updates in a while. It sounded like he allowed you to be a Beta tester.

[traynor]

Originally Posted by whodoyoulike

Haven't seen any updates in a while. It sounded like he allowed you to be a Beta tester.

I don't know what he is doing in the handicapping area, but his approach to data mining (and related data analysis) seems (conceptually) about as good as it gets.

[raybo]

Originally Posted by Capper Al

Under $20 a race.

Betting under $20 per race? Man, there's probably a whole bunch of those.

Now if you change that to making $20 profit per race, long term, you reduce that by gobs. It has been estmiated that 2% or less of all players make ANY net profit long term, so if you can estimate how many players there are, then you can get a better idea - LOL. Good luck with that. By "long term", I would say 8 to 10 years, a couple here would say " lifetime".

[whodoyoulike]

Originally Posted by raybo

If your decision process is not based on "intuition", but concrete data, then yes it could be programmed.

But, does your software program consider this type of situation? Anyone with a black box program can respond. There are numerous situations which come to mind, another is a horse which is tearing up his races for 6f at a "C" track and then is entered at an "A" track similar distance. Do you blindly follow your BB selection without evaluating the selection? How much would you wager? I wouldn't wager on him until I see how he performs.

Several people have indicated they would and everyone should also bet the BB selection(s) because the program selected it. There are too many exceptions in horse racing for me to be comfortable with this approach of no subjectivity.

[raybo]

Originally Posted by traynor

I don't know what he is doing in the handicapping area, but his approach to data mining (and related data analysis) seems (conceptually) about as good as it gets.

I have to agree with that. If he can indeed create a suite of software that can analyze historical and future races, and modify and weight factors, on the fly, he will have taken a huge step towards creating a viable black box. However, there is a lot more to it than just that, as I'm sure anyone who has ever created models will testify. One must constantly maintain the database and continually upgrade and verify the software's processes, for example. User support is a huge part of a project of this scope also. Anytime you put a piece of software out there to the public, free or not, you take on the responsibility for user support functions, and from the sound of what I have heard so far that is going to be significant.

[raybo]

Originally Posted by whodoyoulike

But, does your software program consider this type of situation? Anyone with a black box program can respond. There are numerous situations which come to mind, another is a horse which is tearing up his races for 6f at a "C" track and then is entered at an "A" track similar distance. Do you blindly follow your BB selection without evaluating the selection? How much would you wager? I wouldn't wager on him until I see how he performs.

Several people have indicated they would and everyone should also bet the BB selection(s) because the program selected it. There are too many exceptions in horse racing for me to be comfortable with this approach of no subjectivity.

Since you quoted me, I'll respond. My program does consider distances and surfaces run in the past to address changes in distance, and quality of fields and adjusted velocities (fractional and total) in past races to address changes in track.

Except in very high class races like the TC and BC, and most stakes races with purses of $250k or more (because there are far fewer races of this type, per track, in the database. I have a separate database for each track.), I follow the program's output to the "T", otherwise all my database data means nothing. Regarding betting, I bet the same amount (flat bet), generally based on 1/2% of bankroll and 2% of last cycle profit (updated by card, not race) at most tracks I play, that varies by track, according to how the track tests and which rankings method I choose to use, also according to my track testing activities. However, before my ADWs all dropped me due to Texas state law outlawing internet wagering, I kept my bankroll rather low, and as a result my bets rather low (I no longer wager due to not being able to wager online, and live over 70 miles from the nearest track, and I absolutely hate betting at the track anyway). I don't try to make big profit with the program, which was designed for win betting primarily, rather I use that win profit as cash flow to support my superfecta play, which is higher risk but with much higher reward potential.

My view on "subjective" analysis, is that, in the long term, most players will "guess" wrong more times than they guess right. And that is what subjectivity is, guessing based on prior experience and knowledge, IMO. Lack of discipline and lack of consistency comes into play when your method contains too much subjective analysis. Basically, I believe in the old saying, "Keep It Simple, Stupid".

[traynor]

Originally Posted by whodoyoulike

I'm getting the impression that you are familiar with both recreational and professional types of software. In either type, do you place a wager on every output selection? Does it indicate how much or when not to make a wager?

My software will always provide a best to worse ranking but I use my judgment whether the rankings are valid and worth a bet.

I think a lot of confusion come in when "every race is handicapped" and the software is expected to "make selections" for every race by comparing the entries in that race in some (relatively simplistic) aspect(s) of their performances. "Pace" or "speed" analysis would be an example.

The professional-level software I am familiar with doesn't work like that. Meaning it may generate "selections" in two races, three races, a half dozen races, or no races at all at a given track on any given day. Do I bet on every selection? The simple answer would be "yes" with the addition of "almost always." There are exceptions, but they would take long, boring explanatins to cover.

[raybo]

Originally Posted by traynor

I think a lot of confusion come in when "every race is handicapped" and the software is expected to "make selections" for every race by comparing the entries in that race in some (relatively simplistic) aspect(s) of their performances. "Pace" or "speed" analysis would be an example.

The professional-level software I am familiar with doesn't work like that. Meaning it may generate "selections" in two races, three races, a half dozen races, or no races at all at a given track on any given day. Do I bet on every selection? The simple answer would be "yes" with the addition of "almost always." There are exceptions, but they would take long, boring explanatins to cover.

I agree, a good program will NOT generate "betting" selections in every race, sometimes no races on a card. This simply validates to you that the program is doing what it was designed to do, put emphasis on the best races for wagering. There are times when one makes a conscience decision to play races that the program tells you should not be wagered on. For me, that happens only in the big races, where the pools are huge and potential betting value is very high. On all the other races, those races where the most analysis has been done, because of the relative frequency and numbers of those types of races, I trust the program to point me to good betting opportunities.

[Ed Keenan]

Your summary reply contrasting the utility of decision-making software for the professional vs. the hobbyist, highlights the first-order issues: an "omniscient black box' versus a final decision, based on subjective factors, gets to heart of what the software might be expected to do for the bettor. How far has CX Wong [Precision, copyright 2011] been able to advance this project? Do you or anyone else have any thoughts on far the Author and/or his Asia based syndicate has been able to advance this project?