Profile players

[Tadek]

Hi
I am new at this forum, my name is Tadek and I am a software developer. I am interested in your opinions about profiling of races and betting strategies for different profiles. I am currently designing a module for the next release of my product and my question comes down to: “What features would be useful in a software package for profile players”.

Any thoughts will be appreciated.

[chickenhead]

what kind of profile are you talking about?

[Tadek]

The best way to explain a profile is with an example. Let’s take Woodbine 2003, Races at 1 Mile and above without Maiden Claimers. One evaluation method produces: profit of $88.00 on 364 races with 98 winners (26.92%); the ROI is 0.12, the minimal cash position (“back draft” in some circles) is -$28.20 and the longest loosing streak is 14. That completely defines a set of possibly playable races.
My question is: Do you have a systematic method of playing this kind of a profile. Would you find useful parameters like “Volatility” of a profile (ratio of the profit to the largest payout and the ratio of the profit to the average payout), have you ever used a sliding window strategy or any parametric method? Any ideas will be appreciated.

Thanks for your interest chickenhead.

Regards

Tadek

[garyoz]

Sounds like another model based on back-fitting races to predict the winner. At least that's what it sounds like to me. These "profiles" tend not to hold up when picking future winners. Have you tested it against a large sample that was not used in developing the profile? In terms of parametric measures, there was an extensive thread last year or early this year on parametric and non-parametric measurements. You might want to do a search on these terms.

[ranchwest]

Tadek,

ROI is measured based against 1.00 as break even. I believe you meant to say 1.12 instead of 0.12.

Stated another way, zero income would result in an ROI of 0.00.

[Tadek]

Profile playing is for those who believe in statistics. General assumption is that if a handicapping method applied to a particular set of races produces profits during the current meet there is a chance that this method will remain similarly profitable for the next N of races. The profile I’ve presented above was completely useless for first three months of the meet. In July it showed 0.14 ROI and I’ve decided to play it, it remained profitable for four out of next five months. The same method proved to be disastrous for the 2002 meet at the same track. A similar strategy held superbly well for this year’s Santa Anita and again last year’s meet was useless. In general I don’t believe there is a method, which can produce consistent results on different tracks over many seasons. The challenge is to discover what ‘holds’, play it and bail out before it becomes negative. Now, I am trying to define and compute indicators, which might provide a hint about entry point and possible exit from a profile. Issue of avoiding loosing bets is another interesting challenge.

Thanks to garyoz for useful directions.
Regards

Tadek

[Tadek]

Hi, ranchwest

I have to admit that ROI is confusing to me. My assumption was that ROI is the profit per dollar wagered. I believe I saw it defined that way somewhere. In the profile I’ve presented above the computation of ROI was done this way:
(TotalPayouts – TotalCost)/(NbrOfRaces*SingleWagerCost)
This way loss produces negative value.

Thanks for pointing it out

Regards

Tadek

[garyoz]

Tadek, it is important not to confuse correlation and causation. Do you know why the model held up at Woodbine for the period you mention but not at other times or at other tracks? IMO, you need to have a handle on the underlying factors causing the model to be successful. Not to be too sarcastic, but you know the old adage, even a broken clock is correct twice a day.

There are plenty of handicapping software programs that undertake modeling. BTW, if you choose to answer, what is your source of data, BRIS, ITS, Trackmaster, HDW, DRF, homegrown? Also are you using conventional variables from racing "literature?" Not to sound too much like an academic.

[BeatTheChalk]

The only profile that I have seen...that will keep me off the
losers : Model Per Cent Early Energy. Keep records for your
track and soon you will see which horses CAN NOT WIN..in
certain situations.. races, etc. Many times the Favorite will have
the Wrong Per cent early...i.e. the horse runs too fast early and
can not finish on top. That .. gents..is all I know about profiles.

[andicap]

Originally posted by BeatTheChalk
The only profile that I have seen...that will keep me off the
losers : Model Per Cent Early Energy. Keep records for your
track and soon you will see which horses CAN NOT WIN..in
certain situations.. races, etc. Many times the Favorite will have
the Wrong Per cent early...i.e. the horse runs too fast early and
can not finish on top. That .. gents..is all I know about profiles.

The one problem I havewith %E is that a horse's energy is all over the lot. In one race it could be 51.24 and in the next 53.68. How do you know which %E to use???

[BeatTheChalk]

Your question about a horse...that shows widely varying figures
for " Percent Early." Frankly I dont know ! I use the Aodds
software...that picks pace lines. Whatever that pace line shows..
is what I look at in the readouts.
I use Thoromation for my final decisions. The Percent Early is
displayed clearly for each of the horses. I will eliminate certain
horses IF they simply run too fast Early ( for Today's race )
It works more times than it doesnt work.. but obviously it wont
work all of the time. Most of the longshots that we get .. are
in races where the Favorites -- do NOT FIT the profile that is
necessary to win today.
That is all I can say. I follow Calif , NY, Fla, and Kentucky. I
pretty much know what it takes to win... at different distances.
Also different tracks. Hollywood - Santa Anita and Delmar have
different profiles. Same for the other circuits.
I hope this helps...but it isnt easy .. no matter how much I
think I know :) Here is an off beat example:
I ran the Belmont stakes. I gave Birdstone his win in the
Champaigne stakes as a two year old. Some folks would say
" Hell you cant use that pace line." I said Ok...I gave him his
race in the Kentucky Derby. ( Looks terrible on paper ) Guess
what ? He shows the same in the read outs..no matter which
pace line I used !
Finally: to really do this right .. you have to keep Records..it is
a ton of work. Right now I am not up to it ...So I just do certain
races at certain tracks where I think I know what it takes to win.......LOL...and sometimes It works.

[Tadek]

Gambling literature often mentions two types of players: statistical and probabilistic. Statisticians don’t really care “why”; all that matters is that a series of events produces a positive expectation and we know how to benefit from it. Probabilistic players attempt to model and rationalize factors that produce the result; previous races have no influence on the horse that is about to cross the line. In my opinion finding rational explanation for the results is the biggest challenge we face. We all come from the same root; we love the game and want to know “why”; many times however the answer eludes me and in those situations I will settle for the monetary reward while remaining in obscurity when it comes to causality.

I have developed my system because I wanted something flexible, factors can be added after the system is shipped. Computations are performed by a technology called “Fuzzy logic evidence gathering” – (there is nothing fuzzy about it but computer people love to use sophisticated terminology to describe simple concepts). An evaluation method (handicapping method) combines factors into a tree like structure and then computes combined results; all regression methods are equivalent in the end. User can define his own method by assembling factors. Profile statistics are computed by applying a user selected evaluation method on a set of races. Profiling is done interactively by selecting race categories and then applying a wager to compute results. Factor influence analysis can be done for each profile – this way one can search for answers “why” and also can adjust parameters to optimize for profit. System uses exclusively DRF data exported by the Formulator.

To terminate I will show you two profiles I found at SA this year and I’ve effectively played both of them: Evaluation method was “Speed, Workout and Competitiveness” for both;
Profile A: (no maiden claimers and no claimers)
1st pick – $157.40 profit; 360 races; 97 winners; 27.71% winners; 22 c/$; longest loosing streak 10; minimal cash position + $6.00

Profile B: (maiden claimers and claimers)
2nd pick - $31.60 profit; 332 races; 73 winners; 21.99% winners; 5c/$; lls 25; mcp -$4.40
I’ve tried to find out why the 1st pick in the second profile was badly loosing while second pick was profitable and I was unsuccessful. One of those things.

Thanks to garyoz for discussion, love talking to you.

Regards

Tadek

[acorn54]

i use equisim. that has a profile aspect. from the database nathan provides (56,000 races), the profile report beats the take out, but does not show a positive r.o.i.-guy(acorn)

[garyoz]

Tadek, best of luck on the fuzzy logic approach. Neural networks and fuzzy logic approaches have been used in several programs. For example, BRIS has Neurax and Neurax Pro. I had a brief but interesting discussion with its developer. I think that the Winner's Circle (John Rancourt) was using Fuzzy Logic more than a decade ago. I'm certain that there are other programs. You might glean some insight from their approaches. Hope your work pays off.

[Tadek]

Thanks everyone. I really appreciate your input. It seems that there are not that many profile players out there, oh well, my product hits the market on July 1 and after that I will see.

Good luck everyone

Regards

Tadek

PS
Yesterday at Bay Meadows in the 7th race my 2nd pick beat my 3rd pick by a nose. I was able to back that horse at 48-1, it certainly made my day.

[garyoz]

Oh, so this has been a sales pitch. Well, there's alot of products trying to do what you do. You should have been upfront in your initial posting saying that you were going to be marketing a product. Personally, I could care less about your redboarded $48 winner.

[Tadek]

garyoz,
This is not a sale pitch. I did not give the name of my product and I did not indicate where it can be found. I am really interested in concepts and ideas, which might lead to new features in my system – which I pointed out in my first message. Discussion boards like this one are very strict when it comes to advertisement and I agree with that policy; I don’t like receiving unsolicited adverts and therefore I am not giving any.

Looking at my last message I admit that it might be perceived as an advert, I apologize for it. Yesterday’s win got me a little bit too high and talking about it while stating the date of release of my product was incorrect.
I apologize one more time and I promise that there will be no mention of my software ever again and I will take care to review what I am saying to avoid this kind of embarrassment.

Thanks again to every one and I hope you will not ignore me when I post other subjects.

Regards

Tadek

[chickenhead]

what is your output? An odds line, a ranking, projected finish position, beaten lengths, final time?

[Tadek]

chickenhead
After receiving solid kick in the butt from garyoz I really don’t want to talk about anything relating to my software. I appreciate your time and all of your posts but this thread should end now.

Regards

Tadek

[garyoz]

Asking for input on a software product 13 days before a launch? Not very likely.

[ranchwest]

Originally posted by Tadek
Hi, ranchwest

I have to admit that ROI is confusing to me. My assumption was that ROI is the profit per dollar wagered. I believe I saw it defined that way somewhere. In the profile I’ve presented above the computation of ROI was done this way:
(TotalPayouts – TotalCost)/(NbrOfRaces*SingleWagerCost)
This way loss produces negative value.

Thanks for pointing it out

Regards

Tadek

ROI is Return On Investment. Literally, it is Return/Investment. So, if you win $3 on a $2 wager, your ROI is $3/$2 = 1.50.

[Dave Schwartz]

RW,

Only in the horse racing world.

If you talk to an accountant about your "ROI" he will express it in a form such as "10%" or "minus 3%."


Regards,
Dave Schwartz

[chickenhead]

Tadek,
It's not a problem to talk about your software, just don't try to pitch it. You do understand the difference?

My original question still stands, I'm curious how and why people have chosen to quantify their "picks" the way the do. Question is open to anyone here.

[ranchwest]

Originally posted by Dave Schwartz
RW,

Only in the horse racing world.

If you talk to an accountant about your "ROI" he will express it in a form such as "10%" or "minus 3%."


Regards,
Dave Schwartz

My degree is in accounting. I stand by my statement.

You are describing gain/loss on investment, not ROI.

[Dave Schwartz]

RW,

Really? (Understand I am not being sarcastic here, I am being serious. Please educate me on this subject.)

So, if I talk with an accountant about the appreciation of my net worth last year that was (say) up by 80% he will say it was 180% or 1.80?

And he would express a loss of my net worth not with a negative but rather as .75 or .75%?


Dave

[GameTheory]

What do form do accountants use when talking to other accountants?

[Buckeye]

This sounds like an important question. :cool:

[ranchwest]

Originally posted by Dave Schwartz
RW,

Really? (Understand I am not being sarcastic here, I am being serious. Please educate me on this subject.)

So, if I talk with an accountant about the appreciation of my net worth last year that was (say) up by 80% he will say it was 180% or 1.80?

And he would express a loss of my net worth not with a negative but rather as .75 or .75%?


Dave

Oh, Dave, you aren't going to like this at all. :)

Net worth is something entirely different.

Net worth is assets less liabilities. Net worth is computed as of a particular date (or date and time, if you will). An expression of the change of net worth would be a comparison of one date to another date.

Amounts such as profit/loss or ROI are for a time period -- from one date to another.

An accountant might express values in many different forms, but the question was on ROI.

ROI is "Return on Investment". Return is your proceeds. Investment is your outlay. Return on Investment is the return divided by the investment. It is exactly what it says it is, return on investment.

I believe ROI is most commonly used on projects.

[ranchwest]

Originally posted by GameTheory
What do form do accountants use when talking to other accountants?

Most commonly a 1099 or an invoice. :)

[Secretariat]

I agree with RW, but its kind of a moot point.

If you invest 2.00 and receive back 3.00 your return on (or over) investment is $1.50 for every $1.00 invested. So describing your Return on Investment is Return/Investment or Money Returned/Money Invested. ROI mean Return on (or over)Investment.

Dave, your reference to "appreciation" based on your existing capital doesn't factor in investment.

If you made 40% more money than you did last year, it does not account for the "investment" aspect.

For example, say I made 20,000 last year, and this year I am making 30,000 or 50% more. This may be seen as a 50% increase in my earnings, but not necessarily a return on investment. However, if I invested 20,000 and received 30,000 back then taking Return and dividing by Investment equals 3 dollars return for every 2 dollars invested or 1.50 Return Over Investment or Return on Investment. Because the investment is 1.00, one could say that the 1.00 investment will lead to 50% profits. I think the word "on" rather than "over" investment sometimes causes confusion.

Personally, as long as your users are all in agrement on the way you're calculating or interpreting it, it's kind of a moot point because its a terminology thing, and I think most of us know if we've got 1.5 times more money than we started with. Unfortunately, for most, generally its less. For a negative ROI a 0.88 would mean a Return of 88 cents/Investment 1.00 or 88 cents back from each dollar invested or a 12 cent loss per dollar invested.