2-day special

[Dave Schwartz]

Anyone from the PA board considering a purchase of HSH?

If so, order by Tuesday, November 25th, 2003 and get a $100 discount for mentioning this post.


Dave Schwartz

[COUGAR]

Mr Schwartz..

Unfortunatly i have to many irons in the fire to take you up on your wounderful offer at present. I have heard nothing but the best of your software from this board and i am sure it is one of the premier products on the market today. I look forward to checking it out here in the future as some things clear up with other projects. All the best and a wonderful holidays to all on the pace advantage board...

Cougar

[Dave Schwartz]

Cougar,

Thanks for the kind words.

And to everyone else, one more day. <G>

Dave

[lsbets]

Dave,

Wish I knew about the upcoming offer two weeks ago when I bought HSH. :)

Seriously, for anyone considering buying the program, I am very impressed with it. Its much easier to use than I thought it would be, and I have had some very good luck so far even though I am still trying to figure out the best way to use it. Dave is a tremendous help and brings a ton of knowledge to the table, and the other users on his board offer a wealth of advice.

The only warning I would give anyone is that if you are going to call Dave, set aside at least an hour. When he gets on a roll talking about HSH, you might end up on the phone forever.

[formula_2002]

Dave I just took a brief tour of your web page.
It would appear that you give a factor analysis (example, Post Position) which includes # starts, win % IV, net ROI etc.

Having work with the factors in ALL-WAYS / BRIS data files for several years, I have found that they are extremely accurate in finding the most probable winner, however no one factor beats the track-out. Also I have not found any combination of factors capable of beating the take-out on a significantly statistical level.

Would you say that any one of your factors or combination of can beat the track take-out on a significantly statistical level?

For significance I use Mitchell’s Test for significance. That is show a profit in “N” number of races at small incremental odds ranges where;

N= Win% x loss% x(1.96/.05) x (1.96/.05)
Where win% would allow for say a 10% profit, 95% confidence and 5% error.

Thanks
Joe M

[Dave Schwartz]

Formula,

There is no single factor that I am aware of that beats the track take by itself ACROSS ALL TYPES OF RACES. I feel that anyone that is searching for such a factor is wasting their time.

Of course, that isn't the issue anyway. The real issue is, "What wins a race like this one?"

Interestingly, I made this point in another post just yesterday. Here is the link:

Modern Impact Values (http://www.paceadvantage.com/forum/showthread.php?s=&postid=75607#post75607)

Consider that profitability (or value) in a given race is driven by the public's perception of a horse's chances. Those perceptions are driven by factors. Is it not safe to say that horseplayers in New York have a "somewhat" different value system than those in California? or Wyoming?

In addition, the impact of a factor will certainly change from one "race configuration" to another. (Note: By "configuration" I mean the unique permutation of circuit, track, surface, distance, race type, age, sex, track condition, month-of-year, field size, number of front runners, strength of favorite, etc.)


Let's ask a few good questions.

Do not Quirin's Early Speed Points have a stronger impact in 6fur races than at 7fur? Is it not logical that they would have a different impact at a track with a strong speed bias?

How about speed ratings? Is it not logical that the impact of speed varies from track-to-track depending upon variables as wide ranged as the abilities of the chart callers?

What about the earnings box? Do you suppose that slots have an affect on the accuracy (or inaccuracy) of ratings derived from the earnings box?

This is why racing is so difficult. Every race is almost unique. (I say "almost" because, like beauty, uniqueness is in the eye of the beholder: It depends upon one's viewpoint.)

Statistical guys insist on trying to reduce the game to one giant "study." That NEVER seems to work.

As soon as I mention breaking the races down into these "configurations" the statistician complains that the sample size gets too small to study. So, they'd rather have enough races to study even if those races are so different that they could more easily be labelled "disimilar" than "similar."

The pragmatist says, "I'd rather have a smaller sample of races like this one. At least I will find something that the public has wrong." - And that is where the profit comes from.


So, in closing - My answer? I routinely find profitable (sometimes highly so) factors that apply to a given race. Of course, I don't use THAT approach either. <G>


Regards,
Dave Schwartz

PS: This is the last day of the PA sale.

[formula_2002]

MAGNIFICENT!!!
Take’s everything right out of the equation...


Joe M

[Dave Schwartz]

Formula,

Not sure what that means because we certainly do use "equations," and by the hundreds.

Could you clarify?


Dave

[formula_2002]

Dave, I gather from your reply to my question means that statistical analysis is out of the question.

Joe M

[Dave Schwartz]

Joe,

LOL - Well, I can see which category mentioned above you fall into.

No, statistical analysis is not out of the question. <G> But rigorous statistical endeavors involving giant amounts of data will likely prove futile. (Just as you have seen in your experience.)

Understand that this is not meant as an attack, but I have watched your posts here on PA for about the last year or so. You continually say that what you are doing is not working (and never will) yet are not willing to embrace any other approach.

Think of it this way... We want to study red, somewhat round, firm fruits (generally referred to as apples). The problem is that we don't have enough apples to study. So we include some other fruits that are of "similar" shape and color in order to get a larger sample. Pretty soon we are studying oranges as well as apples. Is it any wonder that we find our study is not "fruitful?" (Sorry, couldn't resist.)


Your statistical knowledge could have great merit, providing you are willing to give up sample size. And to not give up sample size is to guarantee that you will find nothing that works (as you have already stated).

Look for similarities in your data! There are answers to be found!


Best of luck to you.
Dave

[keilan]

Formula -- Statistical analysis is vital, Dave is saying don't throw everything into a mixing bowl and expect to gain any worthwhile conclusions -- cause you won't!

Quote from Dave -- "In addition, the impact of a factor will certainly change from one "race configuration" to another. (Note: By "configuration" I mean the unique permutation of circuit, track, surface, distance, race type, age, sex, track condition, month-of-year, field size, number of front runners, strength of favorite, etc."


I don't always agree so emphatically. Why not take his advice and test it, you have everything to gain.

[formula_2002]

K, just for the record, I analyze the 85 or so factors I follow by race type, age, speed shapes, distance, surface, tracks, race catagory, race types, odds, on and on.

I also determine the win %, IV, ROI for each factor and for hundreds of combinations....all on a database of over 160,000 horses...the more I analyze..the more prone I am not to bet....

Oh, I also test my resulst statistically...

I wish I lived in Dave's world...;)


Joe M

[Dave Schwartz]

Formula,

>>I wish I lived in Dave's world<<

Well, yes, you probably do. <G>

If you ever change your mind I'd be happy to spend some time with you explaining what I believe you are missing from my posts.

Just let me know and we'll set up a time to discuss it.


Dave

[keilan]

Joe -- I've had my say and really don't have anything further to add. Sorry I wasn't able to help.

[formula_2002]

Thanks for the input fellows,, appreciate your time,effort and concern

Joe M