statistical analysis?

[formula_2002]

Does anyone here review their results with statistical analysis?
If so, what level of confidence do you achieve?

I'll mention this. One time when I asked the same question, someone responded saying "they just keep track by counting the Franklins".

Not what I'm looking for. :cool:



Thanks

[Augenj]

Does anyone here review their results with statistical analysis?
If so, what level of confidence do you achieve?

I'll mention this. One time when I asked the same question, someone responded saying "they just keep track by counting the Franklins".

Not what I'm looking for. :cool:



Thanks
Is this what you mean? If not, point me in the right direction.

http://www.paceadvantage.com/forum/showthread.php?t=132894

[sjk]

Years ago you had a link to a page to calculate this. If I input a large set of data there was a large degree of confidence.

[formula_2002]

Is this what you mean? If not, point me in the right direction.

http://www.paceadvantage.com/forum/showthread.php?t=132894

sorry I put my reply's to your thread
http://www.paceadvantage.com/forum/showthread.php?t=132894

[formula_2002]

Years ago you had a link to a page to calculate this. If I input a large set of data there was a large degree of confidence.

I cannot believe how long I've been doing this stuff.
I searched going back 10 years :)

did not find it

If you like I can post an excel file I'm currently using

[Nitro]

Does anyone here review their results with statistical analysis?
If so, what level of confidence do you achieve?

I'll mention this. One time when I asked the same question, someone responded saying "they just keep track by counting the Franklins".

Not what I'm looking for. :cool:
Thanks
My question is WHY just a “statistical analysis?”
Every decent money management system I’ve ever seen (and use) involves just that: $Profit & $Loss. What sort of “results” could be any more important than that?
My comfort zone is not just based on a winning hit frequency, because I’m always looking for value from the chosen selections. Who cares how many times you score? The real substance is in the return value of each score.
So even though the races I pass might provide a winning outcome as far as the runners go, they may not provide the profit margin I’m looking for.

It'll be interesting to see how this thread plays out.

[VigorsTheGrey]

My question is WHY just a “statistical analysis?”
Every decent money management system I’ve ever seen (and use) involves just that: $Profit & $Loss. What sort of “results” could be any more important than that?
My comfort zone is not just based on a winning hit frequency, because I’m always looking for value from the chosen selections. Who cares how many times you score? The real substance is in the return value of each score.
So even though the races I pass might provide a winning outcome as far as the runners go, they may not provide the profit margin I’m looking for.

It'll be interesting to see how this thread plays out.

I was shocked to see the payout for the Pacific Classic...I thought there would be no value if the top selections finished 123...and I wagered trying to get value by using the others in one of the top 3 holes...chasing value...my handicapping and instincts told me Chrome, Beholder, Dortman and then maybe Hoppertunity and Dalmore after the top three.....silly me..top choices run 123. $1 ex pays 6.5 to 1. $1 tri 12 to 1. $1 super 37 to1. How much value does one need? Geez! How dumb can I be?

[AltonKelsey]

Those are the prices FOR $1 not TO $1, so subtract 1 for the real odds.

1-1 over 3-1 third choice, wouldnt be expected to pay much less than that. Hardly a bonanza, unless you thought it was some kind of lock . Dortmund no shot to place? Not really.

Chalk is usually going to pay fairly, as many dollars are wasted chasing supposed 'score' payoffs that are in fact, total ripoffs most of the time when they hit.

[sjk]

I think this is the one; the link to the test no longer works (I am sure you would know if it did as it was your page).

http://www.paceadvantage.com/forum/showthread.php?t=16662

[traynor]

Does anyone here review their results with statistical analysis?
If so, what level of confidence do you achieve?

I'll mention this. One time when I asked the same question, someone responded saying "they just keep track by counting the Franklins".

Not what I'm looking for. :cool:

Thanks
First question, yes.

Second question, wrong question.

The question should be, "How often do the results of the analysis represent values that can be meaningfully applied to the general population?" The answer to that question is, "Almost never."

The reason is simple. "Level of confidence" implies that the results obtained (from specific races) are perfectly representative of the general population (of all races). That is almost never the case--no matter how large the sample, it is NOT (in most cases) an accurate representation of the distribution of events in the general population of all races.

It is not a trivial problem, nor is it one that can be resolved with simplistic "statistical analysis of confidence levels." It is the difference between predictive analytics and descriptive analytics. The latter does not account for the fact that each horse racing event causes a change in the perceptions and behavior of the betting public, when contrasted with the perceptions and behavior of the betting public before the outcomes of the individual events were known.

For a better understanding of the problem, I recommend a thorough review of Schlesinger's works on blackjack:
https://en.wikipedia.org/wiki/Don_Schlesinger

Of particular relevance is Schlesinger's statement in regard to the result of Anderson's three years of full-time blackjack play as "statistically irrelevant." No matter how big the model constructed for horse racing (because the outcome of events alter the outcomes of future events, most significantly in regard to the amounts wagered), it will never be more than an approximation of possible future outcomes.

Seeking a "level of confidence" is admirable in other types of statististical analysis. In horse racing, it is likely to do more harm than good (for a bettor) by implying a degree of certainty in future outcomes that does not exist in the real world.

[Parkview_Pirate]

Of particular relevance is Schlesinger's statement in regard to the result of Anderson's three years of full-time blackjack play as "statistically irrelevant." No matter how big the model constructed for horse racing (because the outcome of events alter the outcomes of future events, most significantly in regard to the amounts wagered), it will never be more than an approximation of possible future outcomes.

Seeking a "level of confidence" is admirable in other types of statististical analysis. In horse racing, it is likely to do more harm than good (for a bettor) by implying a degree of certainty in future outcomes that does not exist in the real world.

So if the predictive analytics of racing are a moving target, then are we not back to closely tracking ROI, as Nitro suggests?

[traynor]

So if the predictive analytics of racing are a moving target, then are we not back to closely tracking ROI, as Nitro suggests?

In the real world, yes. All that matters is winning. It makes no difference whatsoever (at least to me) what the result of woulda coulda shoulda from last year at wherever downs may or may not have been. I have done this long enough to understand clearly that there is no real "level of confidence" that is going to guarantee success.

[formula_2002]

I think this is the one; the link to the test no longer works (I am sure you would know if it did as it was your page).

http://www.paceadvantage.com/forum/showthread.php?t=16662

that is a discontinued link
try my new website
I updated it with my results file which contains the tab "analysis"
which use set up for two odds lines

[formula_2002]

reply to all

The objective in using the statistical analysis is to confirm if one is showing a profit by luck or by skill.


One key here is, if you are beating the odds long term, your betting is Statistically Significant . There is no getting around that.

Just a passing point. If it were not for the concept of "skill"
Fantasy football play would not be allowed in NY.

[traynor]

reply to all

The objective in using the statistical analysis is to confirm if one is showing a profit by luck or by skill.


One key here is, if you are beating the odds long term, your betting is Statistically Significant . There is no getting around that.

Just a passing point. If it were not for the concept of "skill"
Fantasy football play would not be allowed in NY.

Some very knowledgeable people would argue (and have argued) to the contrary--with a lot more acronyms after their names than you or I have. I defer (as I have done in the past) to their superior expertise. However, differences of opinion are a good thing. It keeps the gentle folk betting on losing horses, month after month, year after year, because it can be "mathematically proven" that those horses should have won. I like that. Something to ponder while watching the sunset in Madagascar.

[formula_2002]

Some very knowledgeable people would argue (and have argued) to the contrary--with a lot more acronyms after their names than you or I have. I defer (as I have done in the past) to their superior expertise. However, differences of opinion are a good thing. It keeps the gentle folk betting on losing horses, month after month, year after year, because it can be "mathematically proven" that those horses should have won. I like that. Something to ponder while watching the sunset in Madagascar.

That is exactly why I have not made a serious bet on a horse in years.
I've saved a ton of money because I cannot pick them for a long term profit with a high degree of confidence.

But I do keep looking :cool:

[Nitro]

That is exactly why I have not made a serious bet on a horse in years.
I've saved a ton of money because I cannot pick them for a long term profit with a high degree of confidence.

But I do keep looking :cool:
That’s unfortunate!

You put in all this work day after day and you can’t reap the rewards of all that effort!
I can see now why you’re so concerned about developing a confidence level.
If you were promoting anything other than your tote board studies I would have a lot more to say.

But since you’re a least there you won’t have to be “looking” much longer.
My tote mentor sometimes reminds me that the real basis for his tote analysis formulas are based primarily on human psychology.
GL

[xinyi]

That is exactly why I have not made a serious bet on a horse in years.
I've saved a ton of money because I cannot pick them for a long term profit with a high degree of confidence.

But I do keep looking :cool:See if you can use your statistical methods to show why the old adage 'even a blind squirrel finds a nut every so often' hasn't held in your case. Nothing personal but I don't know whether you 'offend' me more as a horseplayer or as someone who works with data for a living. Good luck

[traynor]

That is exactly why I have not made a serious bet on a horse in years.
I've saved a ton of money because I cannot pick them for a long term profit with a high degree of confidence.

But I do keep looking :cool:

I make serious bets on horses regularly (and MANY more less serious bets) and "do well." It took awhile (in years) to realize what the old investment caveat of "past performance does not guarantee future results" actually meant--and why.

As I wrote earlier, it is not a trivial problem, nor is it one that can be resolved with simple statistics. If I get bored later in the year, I might write a book about it. It is an interesting topic, and one that is as relevant (or more so) in many other fields as it is in horse racing (and blackjack).

"Long-term" profit is tough, if you try to prove it only in reverse (by what happened). The trick (if there is one) is in building predictive models (based on carefully selected subsets of past events), then ruthlessly and relentlessly testing them against events in the real world. Lots of work, but lots of fun. And profitable.

[traynor]

That’s unfortunate!

You put in all this work day after day and you can’t reap the rewards of all that effort!
I can see now why you’re so concerned about developing a confidence level.
If you were promoting anything other than your tote board studies I would have a lot more to say.

But since you’re a least there you won’t have to be “looking” much longer.
My tote mentor sometimes reminds me that the real basis for his tote analysis formulas are based primarily on human psychology.
GL

Probably the most overlooked factor in race analysis.

[Cratos]

I make serious bets on horses regularly (and MANY more less serious bets) and "do well." It took awhile (in years) to realize what the old investment caveat of "past performance does not guarantee future results" actually meant--and why.

As I wrote earlier, it is not a trivial problem, nor is it one that can be resolved with simple statistics. If I get bored later in the year, I might write a book about it. It is an interesting topic, and one that is as relevant (or more so) in many other fields as it is in horse racing (and blackjack).

"Long-term" profit is tough, if you try to prove it only in reverse (by what happened). The trick (if there is one) is in building predictive models (based on carefully selected subsets of past events), then ruthlessly and relentlessly testing them against events in the real world. Lots of work, but lots of fun. And profitable.

Very true and from their pronouncements, very few or if any public handicappers or handicapping books suggest that their handicapping is based on reliable predictive modelling.

[traynor]

Very true and from their pronouncements, very few or if any public handicappers or handicapping books suggest that their handicapping is based on reliable predictive modelling.

Some on this forum have even seemed to take offense at my endless admonitions that there is a (VERY) big difference between descriptive models and prescriptive models. Most anyone with a spreadsheet program and computer can built a descriptive model. Programmers with modest skills (and Anaconda3, for example) can build even better descriptive models. However, they are still only descriptions of subsets of past events. It is in the "predictive" area that computer modeling really gets interesting.

[traynor]

That is exactly why I have not made a serious bet on a horse in years.
I've saved a ton of money because I cannot pick them for a long term profit with a high degree of confidence.

But I do keep looking :cool:

You might want to do a bit of research on:

https://en.wikipedia.org/wiki/Efficient-market_hypothesis

To paraphrase an old NLP maxim, "If all you ever do is the same old same old that everyone else is doing, the only results you will ever get are the same old same old results they have been getting."

For many, horse race analysis is a (tightly closed) information silo. It does not have to be so.

[Nitro]

You might want to do a bit of research on:

https://en.wikipedia.org/wiki/Efficient-market_hypothesis

To paraphrase an old NLP maxim, "If all you ever do is the same old same old that everyone else is doing, the only results you will ever get are the same old same old results they have been getting."

For many, horse race analysis is a (tightly closed) information silo. It does not have to be so.
If I’m not mistaken Formula is already thinking outside the box by not analyzing the horses, but rather the money being wagered on them. As an outsider playing an insider’s game you have to go well beyond analyzing what’s available in the PP’s from any data source. Using the tote opens the door to a new facet of the game that’s overlooked by the majority of players. Many believe that they can determine the outcome of a race without considering the current condition of the animal or the intentions of the connections.

[traynor]

If I’m not mistaken Formula is already thinking outside the box by not analyzing the horses, but rather the money being wagered on them. As an outsider playing an insider’s game you have to go well beyond analyzing what’s available in the PP’s from any data source. Using the tote opens the door to a new facet of the game that’s overlooked by the majority of players. Many believe that they can determine the outcome of a race without considering the current condition of the animal or the intentions of the connections.
Reminds me of the bad guy's line in Liam Neeson's Taken: "Good luck with that." They will need it.