Questions for Mr. Schwartz

[Rune]

I'll be as brief as I can.

1) I haven't actually seen this, so I'd like to have it clarified. Is it true that you posted that you have or had found a combination of factors (I've heard it called an 'ant') that will select 40% winners while playing every race at every track?

2) Why don't you put out a demo version of HSH for prospective users to look at?

3) Why did you make the users' section of your BB private?

4) Why won't you post any selections in advance of the races?

Thanks.

[Dave Schwartz]

Okay, here goes.

>>>1) I haven't actually seen this, so I'd like to have it clarified. Is it true that you posted that you have or had found a combination of factors (I've heard it called an 'ant') that will select 40% winners while playing every race at every track?<<<

At every track I tested in a 5,000 race sample.

>>>2) Why don't you put out a demo version of HSH for prospective users to look at?<<<

Because the learning curve is tough enough that we'd spend too much time on tech support for someone that was not committed to the program.

>>>3) Why did you make the users' section of your BB private? <<<

At the request of our users.

>>>4) Why won't you post any selections in advance of the races? <<<

Because the nature of how I (personally) play is that I make my selections with about 2-3 minutes to post.


Dave Schwartz

[hurrikane]

Dave....5000 races...seems small for auto spot play. How many bets did you get out of 5k races. what was the roi and avg mutual. Can you post some meaningful stats on your play. I have found many 40% plays that show neg return...to much chalk and too many plays.

[Lefty]

I read from Gramps at HTR that there is no peer support
on bbs for HSH. Boy, that's wrong. I need more support
than most everyone and I get plenty. So that's another
point that's wrong.
Peep wants to know why so many bugs. All prgms moving ahead rapidly encounter hugs. It happens at HTR
too. I read the boards. Both programmers quickly quash
the bugs as they are found, so Peeps don't make out
like Hsh only one that encounters bugs.
Thank yuh, thank yuh very much.

[hurrikane]

Dave, maybe I misread the posts above. Areyou saying you played every race every day at every track for 5000 races using the same criteria in every race and got a 40% win rate? Altought that is a little less that 2 mo data it is still an interesting result. Would really like to know about the avg mutual and the roi and any other stats you have avail if you don't mind? Was the play truely blind and was the criteria always the same for every race?

[Dave Schwartz]

Hurrikane,

Specifically, here is what I did:

1. I was working on a prototype of a new genetic algorithm and wanted to test it. The program was designed to play multiple systems (the prototype used 8 internal systems). It would decide which system to use in a given race by looking at a set of rules designed to assign points to each of the 8 systems. After making its way through the rule set for each race it then selected the system with the most points.

2. I grabbed 50,000 races from the database as a training sample, testing on "tournaments" of a few hundred races.

3. Each "Ant Hill" would ultimately put forth one "Ant" as it's representative selector.

4. I trained 5 ant hills for about 2 hours each.

5. The goal of each ant was to simply produce the most winners in its #1 ranked horse. It was not concerned with profitability, just winners.

6. Each ant played ALMOST every race at every track we had in our database (10 circuits at that time). I say "almost" because they skipped races shorter than 5 furlongs, longer than 1 1/2 miles, and hurdles.

7. After training, I then grabbed a new, clean 5,000-race sample for testing. This 5,000-race sample matched the same selection rules as mentioned above.

Then I tested each Hill's "representative" Ant against the 5,000 races. The best hill hit 42% winners. The worst, 34% winners.

Please note that I never considered playing this system... It was designed to see if the training idea would work.



The Controversy
============
When queried about this on the old-old-old PA board I reported that in spite of the 42% winners, that selector lost several cents per wagered dollar. It was pointed out to me that this was not possible.

After looking at the math, I had to admit that the criticizer was correct. That sample could not have produced 42% winners without showing at least around 6% profit.

Understand that when I told about our research I was not making a "claim" for how great our software was. This was not a part of it, just an idea for the future. Frankly, had I known that the system HAD to be profitable, I would have been much more exicted at the time.

As to why the numbers did not add up, I had no idea.

I was pretty much villified as being a charlatan... it was determined by the naysayers that I must be lying.

Recently I located the old code for this software prototype. I spent a little time looking through the code trying to find out how the 42% would have reported no profit.

Actually, there were 2 possibilities: Either the win percentage was reported high or the $net was reported low. (Of course, if you ask the naysayers the third possibility is that I made the whole thing up.)

Anyway, after looking at the code the best I can determine is that ties in the rankings created some problems that might have explained the anomoly.

It appears that the win percentage was, in fact, relatively accurate. .

What the system was intended to do was take the tied horses and sort by LOWEST win price. In other words, it was supposed to prefer the horse with WinPay=$0.00 over a horse with any other win pay.

Unfortunately, it sorted them by FINISH POSITION preferring the LOWEST.

In other words, it broke ties by favoring the correct decision!

Ironically, it grabbed the payoffs themselves in the correct manner. That is, a horse got credited for winning the race, but sometimes got $0.00 for a payoff.

Now, understand that only a small number of races ended in ties (my estimate is around 10-12%). Nevertheless, it does invalidate this stuff for anything that resembles a "test."


SUMMARY
=======
So, when I was asked by "Rune" to elaborate on this, it was a "loaded" question.

I hope I have provided an answer that is "open" enough for anyone to see that there is not now, nor has there ever been a "claim" that our software could live up to these standards. This was simply a software developer saying, "I once built a system that performed tremendously and still lost money." (That was how the thread came about.)


THE FUTURE
========
This prototype was built as a simple test to see if a multi-systematic approach was do-able and could work. I am convinced that it is, in fact, do-able and can work.

Now, more than four years after the original test, we are preparing to release the finished version of this system. (It should be in the hands of our beta testers in the next couple of weeks.)

Will it work? Good question. We shall see.

For more info on this, here is a public thread on our BBS.

http://www.horsestreet.com/ubb/Forum4/HTML/000414.html


Regards,
Dave Schwartz

[Rune]

1) So you found a combination of things (which we'll call a system) that chooses 40% winners over all races "at every track (you) tested in a 5,000 race sample."

1a) Were the tracks tested chosen before the sample was run, or was the sample run, then tracks with unsatisfactory results purged from the sample? In other words, were only tracks which satisfied the high percentage included, or were all tracks originally tested included?

1b) Was the sample run over consecutive days?

1c) Were the same tracks used throughout the sample run?

1d) Did you state, regarding this 40% method, that you didn't know what you had and that you lost it?

2) You say that "the learning curve is tough enough that (you'd) spend too much time on tech support for someone that was not committed to the program."

2a) Does that mean that your stance is that any prospective new users can either take your word that the program is great or not be bothered with?

2b) Why isn't it possible to provide proper documentation to provide most of the support for the software and answer the majority of questions that the user may have?

3) You say that the users requested that the board be made private.

3a) Why couldn't you make the sections of the board that answer tech support questions and/or provide user experiences public so that prospective users could visit it, while keeping sections that members consider sensitive private?

4) You state that you make your plays 2-3 minutes prior to post, which is reasonable. However, not all users play the races in such a manner. If you want to wait until 2-3 minutes prior to post, it's probably safe to say that you do so because you're interested in the odds, gimmick payouts, etc.

4a) Why couldn't you post plays with odds requirements, minimum or maximum (for example, the horse must get off at no less than 5-1 or no more than 10-1)?

And moving on:

5) Was HSH your own idea, or is it a collaboration of several people? I'm not referring to responding to suggestions from users for features. The reference is to the original program itself.

6) Do you do the programming of HSH yourself, or is it done by somebody else to your specifications?

7) Is it true that you stated that one of the reasons that you didn't have HSH pick horses "in real time" for all to view was because a clock was broken and you were afraid that the posts would appear to have occurred after the race was run? Was it your computer's clock? The webhost/server's clock? Has the clock been syncrhonized since?

8) Have you stated on occasion that you've fogotten what certain 'buttons' in the HSH program do?

9) Is HSH your first race handicapping software effort, or have you written (or collaborated on) other (earlier) software of this nature?

Thanks for your time.

[Rune]

Originally posted by Dave Schwartz
Hurrikane,


So, when I was asked by "Rune" to elaborate on this, it was a "loaded" question.


Dave: aren't you out of line (again)? You're assuming that I'm loading questions to trap you. I'm not. I'm asking hard, pointed questions that others who have followed this might be thinking about. There are so many apparent inconsistencies, and since I was offered the chance to ask in a civil, courteous and mature manner, I'm doing so here. This is my opportunity to ask, and this is also your chance to clarify things, if you so desire.

Excuse me for not asking nice questions or whole-heartedly endorsing your program as the next coming of the Messiah. I'm old-fashioned. I like to see some semblance of proof before I believe.

[hurrikane]

Dave,
I have talked to you on the phone a couple of times. You have some interesting ideas.

One thing I'm wondering...why do you say at 40% winners it couldn't lose money? How was it explained to you.

[Tom]

You know, a guy writes a program and tries to improve it, and is willing to share some ideas with us and he gets nailed to the cross for it.
If you don't trust Dave of his program, don't buy it!
Real simple concept. I have been to his board any times and I read a lot of people who seem pretty satisfied with it. I considered it myself for a long time, but I went to HTR simply because it was available to me at the time and I would have had to wait for several months for my tax rebate. I still buys Dave's magazine and used to buy his older newsletter and pars from the Thoroughbrain days. He has always been a reputable guy in my book and I for one appreciate his posts and humor.
I know Dick Schmidt personally from Sartin and Pizzolla seminars and I respect him as a person and a handicapper. I know Doc Sartin personally the same way, and he may be a bity off the wall, but the bottom line is I made money with every program he ever sold me (yeah, I used to joke about the program of the month too, but no one held a gun to my head to buy them. I'm a big boy-I can ake care of myself).
I don't understand everyone wanting to bash these guys-what did they ever do deserve this?
(OK, I still can't wait to get home tonight and get on Yahoo to see what they have to say-they will love this thread-it's like driving by an accident-you have to look!).
Anyhow, I've babbled enough - may if we all just listened to each other we might learn something.
I sure don't know it all yet (although I am leading the Beer and Bacon challenge at this point - shameless self gratification on my part HEHEHE).
Dave. Dick. From me, thanks for your contributions, and I don't work for either of these guys.

Tom 526

[Dave Schwartz]

Hurrikane,

>>>One thing I'm wondering...why do you say at 40% winners it couldn't lose money? How was it explained to you.<<<

SImply put, the math doesn't work. Since we were playing consecutive races at all the tracks in our database, it was highly unlikely that, taking average payoffs into account that you could hit 42% winners and not show some small profit. When the math was shown to me, I had to agree.


Regards,
Dave Schwartz

[Dave Schwartz]

Rune,

So, you've taken the time to create this list of questions...

When people ask me questions about my software because they have an honest interest, I respond. I do not believe that you fall into that category.

Guess what? I choose not to be on trial.


SIncerely,
Dave Schwartz

[Rune]

I'm not suprised at all.

[karlskorner]

Dave;

When are you moving back to Miami ?

We'll go to Calder, have a nice bowl of Clam Chowder, a Boston Creme doughnut and coffee from Dunkin Doughnut, stroll down to the paddock, schmooze with a couple of trainers and jock agents, check the shoe board, check the horsey's for changes and appearance, take the escalator up to the 4th floor out to the terrace, catch some rays, watch the horsey's run, go to the window and collect some money, take the escalator back down to the paddock, etc. etc. Life is good.

Between here and Yahoo board, who needs it ?

Karl

[Dave Schwartz]

Karl,

Actually Florida was a strong possibility until a couple of days ago (when we put in an offer on a new house here).

One of our neighbors (3 houses away) moved to Boca in May which opened my wife to the idea. It could still happen.

What are the internet wagering rules down there? Can you use YouBet? Philly Park? Ladbroke?

I know we had this conversation awhile back, so please forgive m,y memory... where abouts do you live down there?


Regards,
Dave Schwartz

PS: Do they have Yahoo in Miami? <G>

[hurrikane]

Dave, I appreciate what you are saying...I'm saying...what is the math?

[Dave Schwartz]

Hurrikane,

Sorry, did not mean to ignore the question. Simply, I do not recall the specifics, but essentially, it entailed assuming that the 42% winners were all the lowest 42% of all payoffs. Even that would have resulted in SOME profit.

Sorry, but that is the best I can do.

Dave

[TonyK@HSH]

Rune,

Just read your post (inquisition) and find it hard to understand the tone. I don't belive I know you and I have no idea if you've had an dealings with Dave. My hunch is you haven't, because I don't know anyone that has dealt with Dave that would share your suspicion.

Since Dave has taken the high road (as usual) and chose not to continue this conversation, let me chime and and give you the viewpoint of a client, long-time product user and now (proudly) a friend!! After all, the testimony of a user should be the real test.

My first contact with Dave came in the early '90's when I called and spoke to him about a Neural Network he was working on. It was the first time we met and he spent literally a few hours on the phone with me talking about his ideas and product line. I personally know several other users that share similar experiences. Dave will not short-change anyone that shows an interest in his products or horses in general.

Since that time I've been a customer through a few design changes (Thorobrain, HNB and now HSH) and know that many of the ideas came directly from Dave. The were some 'tough times' between some of these transitions and Dave had many opportunities to take advantage of me. He never did. During these transitions, Dave regularily ran tests on new ideas which produced some interesting results. I can't remember a case where these results could not be verified by his users. I've managed to win with each of these products and I continue to be amazed at his imagination when it comes to attacking this tough business.

His products are not designed to be a black box. Instead they have always been flexible enough to allow each user to develop his/her own approach. Dave freely shares ideas (as do many users) and there is always new food for thought.

**We may not all agree with Dave's business philosophy (no demo), but he certainly has the right to conduct business as he sees fit. If we don't like it, we are free to shop elsewhere.

**We may not all like the fact that documentation is light with HSH, but his users have agreed that he should focus on coding instead of docs at this time.

**We may not all like the fact that Dave made a portion of his BBS private at the request of his clients, but once again he is taking care of his clients needs.

**We may not all like the fact that Dave insists on getting paid for his work...but remember, most of us insist on getting paid for our time at work as well.

If one looks hard enough, there may be a few things one may not like about Dave's business. BUT, I don't care how hard one looks, there is nothing not to like about Dave and his ethics. He is a fair, honest and good man that deserves better than innuendos. Feel free to make your own judgement about HSH but please do not question his integrity.

Tony

[Ed Hutchens]

If he's a fair honest and good man, why doesn't he answer the questions? It doesn't make sense to me.

[Dave Schwartz]

Ed,

What would you like to know?


Regards,
Dave Schwartz

[karlskorner]

Dave;

Sorry I can't answer your questions about Youbet etc. as I have never used them. I do know that CRC is open everyday until midnight taking wagers from the west coast and overseas. I must admit I am completely ignorant when it comes to simulcast and offshore wagering, since I have no use for them.

As I have stated before, I am in a strictly cash business, I give them cash and they give me back cash, there is no way I can wager "blind" into a track, trainers, jockeys or horses I know nothing about and the last thing I would trust is my computer, I question every line it prints out.

I live in Davie, 7 miles from CRC and 14 miles from GP. It's a good life Dave.

Karl

[Dave Schwartz]

Karl,

Yes, I bet Davie is a good life. LOL - When I was growing up in... well, it was between Hollywood and Miramar (near Pembroke Pines), eventually became Miramar, Davie was "cowboy country."

Grown up quite a bit since, I am sure.

Dave

[Ed Hutchens]

Originally posted by Dave Schwartz
Ed,

What would you like to know?


Regards,
Dave Schwartz

Much thanks for the offer, but there's no reason for it now. I didn't see TonyK's post before I made mine, and he seems to have answered almost every one of Rune's questions for you. It must be nice to have so many loyal friends and customers who come out of the woodwork to support you when you need them.

But the final word with me is that hsh seems to be in the late alpha and early beta stages. I can appreciate the users wanting you to keep working on the program instead of writing help files, but that doesn't help the guys just starting out to learn it any faster. I can understand how you might not know what every button does, because I write software myself and it's easy to get lost in the code. But I just can't justify spending $500 dollars on a program that's in beta development. Perhaps when it's finished, the bugs are gone, and you're caught up with everything. Even if it costs more then.

Much thanks again for the offer to answer questions, but there are too many loose ends at this point, and you're probably too busy to need to waste your time on a person whose mind is already made up.

[PaceAdvantage]

Late alpha, early beta? I have a copy of the program (not the latest version mind you), and I can assure you, it is way past beta. I too am also a programmer with a Masters in CIS.....so I would think that I know if a program acted like a beta version.

How can someone who never laid eyes on a particular program pass judgement that it is a late alpha, early beta copy??? Absolutely amazing.....


==PA

[Rune]

Dave: I don't know about YouBet, because various tracks and corporations that own tracks have had problems with YouBet in the past, but I know that there are no restrictions in Florida regarding Philadelphia Park. Anything that they carrry can be bet on through them from Florida.

[PaceAdvantage]

Wow, four new members in the last day or two...that could be a record. Wonder if this thread has anything to do with it.....nah......


==PA

[Rune]

It's nice being the owner of the forum, eh?

[PaceAdvantage]

Actually, anyone can see who's joined recently. Just click on the members button, and sort by date joined. This function is available to one and all.

And since you're asking....yes, as Amex says, ownership has its privileges.


==PA

[Dick Schmidt]

Damn, leave town for a week and a full blown software pissing match breaks out without me even throwing gas on the fire. Just got off the plane from Hawaii and everyone is picking on poor Dave. Over on the HSH board, that's my job.

Rune, since I'm still on Hawaii time and don't give a damn who I piss off, I'll answer your questions, even the loaded ones. It'll give me a chance to work off all the cranky from being trapped in an airplane for 7 hours. I'm not Dave Schwartz, but I stayed at his house once. (Also, I helped do a lot of this stuff, or at least watched over Dave's shoulder).





[B]1) So you found a combination of things (which we'll call a system) that chooses 40% winners over all races "at every track (you) tested in a 5,000 race sample."

1a) Were the tracks tested chosen before the sample was run, or was the sample run, then tracks with unsatisfactory results purged from the sample? In other words, were only tracks which satisfied the high percentage included, or were all tracks originally tested included?

Tracks chosen first. There are 10 circuits in the database and all were used. Nothing was excluded. We weren't trying to impress anyone (least of all you) but were doing research. Since we actually bet based on this research, what good would cheating do?


1b) Was the sample run over consecutive days?

yes

1c) Were the same tracks used throughout the sample run?

same circuits. Sometimes a track might close and another open, such as SA closing and HOL coming on line the next day.


1d) Did you state, regarding this 40% method, that you didn't know what you had and that you lost it?

No. Dave never loses anything. He has code he wrote for the Apple II in 1983. He said he came across it lately and reviewed the results.


2) You say that "the learning curve is tough enough that (you'd) spend too much time on tech support for someone that was not committed to the program."

2a) Does that mean that your stance is that any prospective new users can either take your word that the program is great or not be bothered with?


This is a "Yes or no, have you stopped beating your wife yet?" kind of question. Very loaded language, I can see why Dave would blow you off. The truth is that each new user seems to require 5 to 10 hours of Dave's time to get up and going. Freebies would require the same amount of time, yet most would not buy. If you question this business model, I would refer you to Microsoft. Ask if they give out free samples of Windows or Office. It comes down to a matter of time. Dave is a one man operation and just cannot support those who are unwilling to pay for his time.


2b) Why isn't it possible to provide proper documentation to provide most of the support for the software and answer the majority of questions that the user may have?

We do provide docs for most questions. It could be arranged a bit better, but it is there. Eventually I'll write a real manual to cover the startup, but even that will take at least 100 hours. Just how much extra are you willing to pay for the program for me to do that?


3) You say that the users requested that the board be made private.

3a) Why couldn't you make the sections of the board that answer tech support questions and/or provide user experiences public so that prospective users could visit it, while keeping sections that members consider sensitive private?

Obviously you haven't even looked at the HSH BBS, as that is exactly how it is set up. Only the user "Picks" section is private, the rest of the board is open to all. You'll find you get a lot more respect and answers if you take the trouble to look before you gripe.


4) You state that you make your plays 2-3 minutes prior to post, which is reasonable. However, not all users play the races in such a manner. If you want to wait until 2-3 minutes prior to post, it's probably safe to say that you do so because you're interested in the odds, gimmick payouts, etc.

4a) Why couldn't you post plays with odds requirements, minimum or maximum (for example, the horse must get off at no less than 5-1 or no more than 10-1)?

Because we look at the odds on all the horses, not just one. And what is the deal with picks??? Anyone who would bet someone else's horses is a scum-sucking bottom feeder who lacks any integrity. If you want to make the bet, do the work!

And moving on:

5) Was HSH your own idea, or is it a collaboration of several people? I'm not referring to responding to suggestions from users for features. The reference is to the original program itself.

Trust me, no one else is twisted enough to come up with some of this stuff. Totally Dave's original.


6) Do you do the programming of HSH yourself, or is it done by somebody else to your specifications?

This question shows you don't have much of an understanding of the economics of software. The only large handicapping program that I am aware of that was "farmed out" is Synergism 6. It took two years and cost about $250,000. This cost will never be recovered. Dave writes every line of code for all his programs.


7) Is it true that you stated that one of the reasons that you didn't have HSH pick horses "in real time" for all to view was because a clock was broken and you were afraid that the posts would appear to have occurred after the race was run? Was it your computer's clock? The webhost/server's clock? Has the clock been syncrhonized (sic) since?

Ya, that was a problem. I have no idea if it is fixed. You already know my opinion on picks!


8) Have you stated on occasion that you've forgotten what certain 'buttons' in the HSH program do?

Buttons! Hell, there are whole sections of the program I haven't even looked at. This sucker is large! I'm sure Dave has forgotten what some buttons do. Of course, he could read the code and find out, but if it isn't important, why bother? Sort of like asking an author if he can quote every line in a book he wrote two years ago.


9) Is HSH your first race handicapping software effort, or have you written (or collaborated on) other (earlier) software of this nature?

No, this is the fourth or fifth in the past 15 years.


Well, that's the way I sees it. I think Dave would answer about the same way if he had the time and inclination. Note to Rune: if you are truly seeking information, you would do well to ask politely and lose the attitude. No one, least of all Dave, owes you anything.

Dick

[Tom]

Originally posted by Dick Schmidt



And what is the deal with picks??? Anyone who would bet someone else's horses is a scum-sucking bottom feeder who lacks any integrity. If you want to make the bet, do the work!



Dick [/B]

Nice to have you back. sounds like you are all rested up and feeling fine after your trip to the islands.
~G~

Very well put!

Tom 528

[crowsage]

Dick Schmidt,

You say this is the 4th or 5th program in the last 15 years. What happened to the others?

Also, you're use of "we" seems to suggest that you possibly have a financial interest in HSH. Do you?

[Dave Schwartz]

Dick,

A correction: Actually, they are genetic algorithms. I don't write neural networks any more.


Dave

[Dick Schmidt]

Crowsage,

I have no financial participation other than as a user.

As for the other programs, the Thorobrain line of several programs is now being sold by another company because Dave was silly enough to trust a lawyer. The Handicapper's Notebook was a database program that eventually leas into the current HSH.

Dick