Profit/ROI

[Milfredo]

Hi all,
Brand new member, but I've been lurking here for some time.

I've been handicapping for quite sometime and have some computer programming skills. I've developed a program(not for sale) which has given me some modicum amount of success.

I've read about every major work on handicapping and have some of the commercial software.

Here's the question......or maybe the problem...I can break even for the most part and with my latest programming effort, am a little ahead.....short term so far. But cannot for the life of me figure out how anybody is making any money, unless it's through the shotgun effect playing exotics. Or if you happen to be Mr. North Dakota and are wired to system.

Any input would be appreciated. I've looked at about every configuration possible, except the brown horse with one short leg and the map of texas in white on his ass.

Milfredo

[Larry Hamilton]

I am not going to say which handicapping service I use, most of you know anyways, but I will say this, if you are using the same data as everyone else (beaten lenths, time at each call, distance, mph, fps, wind velocity, etc) and you are using this data in the same manner as everyone else, why would you expect to win? There is only one set of data, but there is a new way to look at it--composites. Perhaps, this commentary was too vague.

Here are 2 questions I would be prepared to answer.

1) What minimum ROI will I accept?

2) What minimum win rate can I tolerate?

There is a guy at another site who claims a winning ROI while having a win rate that is LESS than random selection (approximately 12%). AS I recall his win rate was close to 5%. I have forgotten the equation that predicts the number of losers he must tolerate but I suspect it would not be uncommon to miss 100 in a sequence. You could stave to death with such a positive ROI. This is why both questions are important.


The answers to these 2 questions then become standards. Everything you do in the handicapping game should focus on achieving what you have set as your standard. ONce those standards have been achieved, raise them. This may sound simplistic, but how many times have you found yourself in the ticket line to buy a straight exacta and come back to the table with a fist full of tickets--you are UNFOCUSED.

Anyways, i expect this conversation will be an interesting one which can explore many facets. I will be delighted to hear them.

Waxing Philosophical: if you assume that it can't be done because you haven't done it (or thought of it) the problem is yours and the solution is find a new game. Question it, Measure it, Test it, but don't deny it until you can prove it ain't so.

[Druther]

You’ve skipped the one winning combination – it’s the brown horse with one short leg and the map of Texas in white on his ass that produces the profits.

Seriously, if you’ve written a program that breaks even, stay with it. You’ve accomplished something I’ve never seen any program accomplish – pick a higher percent of winners and/or lose less than favorites lose. And the reason programs can’t do better than the public can – in my opinion – be summarized as follows:

1.
They’re basically speed or pace programs. They’re too one-dimensional. There are dozens of other factors – such as condition and demonstrated unsuitability to the surface/distance – which are given no consideration in their calculations.

2.
They’re written for the masses. There are many races that are simply god-awful. But the last thing the typical punter wants to see on his screen is a message that says, “This race is indecipherable” or “The odds show there’s no value in this race.”

If I could program – and I can’t – I’d try to incorporate the 2 nebulous factors: Condition and demonstrated suitability/unsuitability to the race conditions.

If one horse has a record of 3-3-0-0 at the distance while another shows 11-0-1-2, it should be factored in.

If a horse won his last race and has demonstrated it can’t win two in a row, it should be factored in.

If a horse earned his highest ever speed rating in his last race – and he’s not a young, lightly raced horse – it should be factored in.

If a horse regularly earns a speed rating in the 70s and has demonstrated it consistently bounces when it earns 80+ it should be factored in.

If a horse only wins when it gets an easy lead it should be factored in.

If a horse has demonstrated that it never wins when returning in less than 14 days or more than 30 days, it should be factored in.

The list is endless. But when it comes to commercially available programs, it’s virgin territory.

[Druther]

Consider, for example, last Sunday’s Beaumont at Keeneland.

The favorite at 3-5 was Raging Fever. Her lifetime record was 7-5-1-0 and her record at the distance was 1-0-1-0. Not bad, but she still didn’t win her lone race at the distance.

On the other hand, Xtra Heat has a lifetime record of 12-11-0-0 and her record at the distance is 3-3-0-0. She’s also the only speed in the race. Her Q is 8. Her nearest competitor has a 4. And she went off at 3-1. And she’s now 4 for 4 at the distance while Raging Fever is 2-0-1-1.

So why did the public – and probably most programs – make Raging Fever the favorite? Probably because in her last race – which was at today’s distance – she earned a higher speed rating than Xtra Heat ever earned in her 3 starts at the distance.

Xtra Heat is sort of like Hallowed Dreams. Neither seems to be viewed as having proven anything. The only thing special about them is they just go out there and get the job done.

[Milfredo]

Hi guys,
Great to see some response. To answer a question posed in general basicaly, my program does use a data file, but I handicap based on what I call style. Hard to explain, but not as number driven as most programs.
I have matched my results against most of the programs out in the public domain and I continually have better results with the three finishers in my top four choices than they do.
One problem, since not as number driven as most, having trouble figuring out how to create an odds line, so I can't look for overlays. I also haven't figured out how to or when to pass races or when a favorite is a false favorite.
So for now, I'm playing almost every race, unless there are 3 or more first time starters. I do require my top selection to go off at 3-1 or better or no win bet, but make an exacta bet every race played.
Any more ideas would be appreciated. Right now hitting on average 3 to 4 exactas out of each 9 races. Key horse with 3 support horses in exactas.

Milfredo

[andicap]

Druther, that was one killer response! Best I've ever seen to a question about software. And I agree with most of your responses.
Unfortunately most people using software don't want to use any judgement.

Milfredo: Try this one: Take your top 3 or 4 choices and only bet those who are 8-1 (or 6-1, try both ways) on the ML. I got this from the guy who programs A-Plus software who swears by it. (this formula is on his website).
I've never tried it, but you might give it a go.
You'll lose a lot of races but hit some big ones: At least that's the theory. If it works, tell me!

andicap

[GR1@HTR]

Humble opinion from one who doesn’t want to use any judgment and who uses using the
same data as everyone else (beaten lengths, time at each call, distance, mph, fps, wind velocity, etc)....

Milfredo, If you have not already, set up an access dB and query how your factors are performing from a win% and ROI standpoint. Get Barry Meadow’s Money Secrets at the Race track. He will explain how to create an odds line and so forth.

Create a chart on how your factors do in...
1) Sprints
2) Routes
3) Non Maiden
4) Maiden
create odds line for races above sorted by field size. ie <=6, 7 to 8, >=9.

If I were you, I would specialize on your best factor and play only those on a very selective basis.

The factor I focus on provides an ROI of -8% to -9% in every race. If I am average, I will lose only 8 to 9% of my money wager. If I am better, only have to make up 8-9 points to break even. Sure in the hell beats trying to make up 17-23 points on a per race basis. Playing only those is like running a 100 yard dash with a 10 yard head start vs at the starting line with every one else.

Ok, that is my take...enough for now.

[Lefty]

Once a racing luminary sold a system for over $300 that
went something like bet the Chestnut horse that goes off at 10-1 or more. This guys still around and sells his
picks.
Making money at the races is tough, ibe glad, or everyone would be doing it and then noone would make
money. I think it's a constant game of monitering and
record keeping, things most hate. I suspect correct use
of a db gives one an edge. Don't know as I'm db illiterate.

[Milfredo]

Very interesting ideas. Prior to getting to this point I have made a line before, based around Mitchel's way of creating one in his book. But I was much more number oriented then.
I've read several places about "Access database"
or importing into excell, but not sure what Access is or how to set up a spreadsheet to import data into.
I just have microsoft works on my computer, no excell.
I believe in keeping records, but haven't had much luck with the programs that keep records and then allow you to train track profiles. I tried pimlico yesterday and matched it up with stk2002 picks that have been trained for Pimlico. I hit three more winners than the stk program hit. Realize that is just one day but still not good enough to make much money because of the low pays on horses hit. Didn't bet, just checked results and wouldn't have made two of the win bets that hit because the top selection of mine was off at less than 3-1. Point is that I haven't experienced much luck playing with trained profiles with any program...same results with All-ways program.
Any help with spreadsheet tips or ideas would be greatly appreciated.

Milfredo

[GR1@HTR]

Think track by track profiles are good and bad. Bad is that it takes a whole meet (1yr) to get only 300 to 400 races under the belt. Personally, I don't think good decisions/conclusions can be made on 400 races much less 5000 races. Good is that you might be able to find some quirky track bias.

[jackad]

I don't remember where I came across it, but I 've been working under the assumption that conclusions (about win frequency, ROI, etc) based on 300 races should hold up over the long term with an accuracy of plus or minus 5%.

What do you think?

[NNMan]

There is an equation that computes the minimum number of races (results) required to support a certain confidence level that those results are repeatable. I think I
originally picked it up in one of Dick Mitchell's books:

Basic Confidence Equation: N = W * L * (Z/E)**2

Where N equals the minimum number of races required for the statistical confidence check,
W equals the win percentage, L equals the loss percentage, Z is a constant depending on the desired confidence level (here we will use 2.575 denoting a 99% confidence level), and finally E equals the maximum allowable error in our win percentage. Here, we will set E to represent a win percentage error that would yield a minimum profit of .10 on the dollar.

I think it works pretty well in letting you know how statistical confidence works in relation to horse race investing. You can go to http://www.handi2k.com/confidence.htm to see a working example.

To Milfredo:

The fact that you have reached "break even" is a monumental accomplishment that you should be proud of. In fact, you can now cross over to the black by simply signing on to one of the rebate programs offered. You should probably run the confidence analysis on your data first before breaking out the bankroll. And this leads to my most important point. If you are serious about earning consistent profit through horse race investing, it must be approached in a strict business-like manner. That means keeping copious records and performing regular, detailed analysis on those records. Boring stuff, I know, but absolutely necessary in keeping you in the black. You must constantly know what's working and what's not and make the necessary adjustments in your investing. I liken it to trying to keep a small Cessna 175 aloft through a string of never ending turbulent weather patterns...

Cheers,

[Druther]

Milfredo

I noticed that you mentioned you attempted making a line along Dick Mitchell’s outlines. Since you stated that you can program, have you written a “capture program” to verify how it performs according to Mitchell’s method?

Now, this is in no way a commentary on Mitchell. The fact is I’ve never used his program (All-In-One) and have absolutely no basis on which to form an opinion. And I have no doubt that some users like what it provides them and that some make a profit with it.

But I do know that All-In-One produces both selections and an odds line. But I’ve never seen any evidence on Mitchell’s web site that his program does what he recommends – automatically tracks his own odds line validity. Why not? What he says – in effect – is that horses you make 2-1 should win “X percent” of those races, right? Ergo, if his program makes an odds line it should also track it’s own performance – otherwise, how would a user know whether or not the program’s odds line is reliable?

Again, this is in no way meant as a critique of Mitchell. It’s more of a critique of programs, specifically, programs that provide you with an odds line.

[stuball]

Have another important factor to consider is how do you bet? Win - Place - Show
Exactas - Tris - Supers.....Many different ways to invest...Keep records..
My personal approach is to bet multiples in the exacta pools using a pyramid
system I have developed over time. If I feel strongly about a race and everything
comes together for me for a certain horse.(meaning race, pace, trainer, odds, etc.)
example: my top horse being the 1 horse -- 2nd choice 2 horse -- 3rd choice 3 horse



$8 ------ 1/2 (1 with 2)

$5 ------ 1 , 2 (1 , 2 box)

$5 ------ 1/ 2 , 3 (1 with 2 , 3)

Bet exactas: $2 ------ 1/all (1 with 2 , 3 , 4, 5, 6, 7, 8, 9, 10)

total bet cost ----$ 46.00

If top choice wins and 2nd choice is second I have nailed the race and have $20.00 exacta.
If top choice wins and anything else comes second I have protected my bankroll and
reduced the aggravation of having the winner and not having the second horse.

If odds warrant I will back wheel my top horse in second.(odds must be sufficient to cover
additional risk of capital)

Bet no races when top two choices pay less than $18.00 (due to low odds or low number of
horses in race.)

Very seldom have a signer (no paper trail) profit is not profit if it is taxed.
Variations due to circumstances but pyramid remains intact.
patience is a necessary virtue for a pari mutuel investor..
Keep records to find strong suits.(truck drivers can not do brain surgery)I know cause
I am a retired truck driver........

Stuball
Have a nice day !!

[Milfredo]

Hey,
Right now I'm betting $10 to win on top choice when 3-1 or better. Then exacta bet is 1 on top of 2 & 3 then 2,3,4 top of 1. $10.00 total on exactas.

I've been playing on weekends for for a couple of months playing a couple of tracks per day. Again playing every race almost anyway because of first time starters.
Right now I'm sitting at almost even....$2.50 ahead I believe.
Very frustrating, as I get quite a bit ahead and then lose it back, then drop down some ....win that back...
must say that I do find it boring to try and keep all the stats as I work during the week 10 to 12 hours a day, and have spent most of my time trying to program and get it right, horse selections and all.
One note, I didn't say I was a great programmer..just some skills enough to have fun and make a rudimentary program that is doing fairly well now. I planned on putting a record keeping section in. I have the headers set up, just haven't taken the time to write the code.
Next move is figuring out how to figure odds line and the real need is to be able to spot false favorites. I figure that is a step that could put some black ink on the page.
One problem is that my 2nd and 3rd choices hit as much as my top choice. Once I figure out how to discern between the 4 top horses, hopefully, by throwing out the false favorite, and looking for overlays...I'll be on the trail.
Any thoughts on false favorites and how to find them?

Thanks guys.

Milfredo

[Dave Schwartz]

Milfredo,

Just a suggestion...

Check out the performance of different odds ranges as your 1st, 2nd and 3rd choices.

You may find that the reason your 3rd choice wins as often as your 1st choice is that there is the boost of having some extra low-priced horses.

If that is the case, you may want to "add some points" for how the public bets the money. This could be because you are not using some significant factor that the public uses.

Just an idea.

[Lefty]

My fav. false fav angle is looking for fav. taking huge drop in class: 50% or more. They do not win their fair share.

[Druther]

I’m almost reluctant to respond because I’m afraid to say that it involves “art.” It requires that you not only know the numbers but how those numbers were earned.

Back in the days when I used the DRF I was always frustrated because – due to space requirements – they provided abbreviated commentary lines. And one of the things I expect from a program is an extended visibility.

For example, here are the actual comments from Equibase for last Sunday’s Beaumont at Keeneland – in order as they finished:

3 Xtra Heat: gained lead, 2-3 path, widened into stretch, kept to task
2 Mountain Bird: stumbled start,settled inside,3-4w bid, loomed, good try,2ndbest
1 Raging Fever: moved up inside,check 9/16s pl,moved outside winner, bid, tired
5 Sweet Nanette: outsprinted early, 5w most of way, failed to seriously menace
4 Carafe: tracked leaders 4-5wide, tired leaving turn

Anyone looking at Mountain Bird’s PP, for example, might well conclude that she not only ran a pretty good race – at 25-1 no less – but that if she hadn’t stumbled her numbers might be even better.

[karlskorner]

Druther;

What would have sent flares up for me with Mountain Bird is the comment "2nd best", surprising how many times a horse with that comment goes on to win next race.

Equibase publishes "Guide to Race Comments" (about 15 pages) that is their "official" guide to chart callers.
You can find it at Equibase under "race products" and print it out, although most comments are understandable, some have very definite meanings. I refer to it often.

Sidenote: I was all excited when DRF started publishing free "charts" on 3/25. ABout 10 days later they stopped putting in the comments, which caused me to e-mail and ask why. My answer was "thats all the information the track (Hialeah) was sending them". Soooo I check other tracks (Aqu, Pim etc.) and found they also did not have comments, I questioned. I am still waiting for an answer. Back to Equibase.

Karl

[Milfredo]

Thanks guys for the insights.

Dave, you're right. I need to have a better idea of what odds reside in general at the different levels.
Again, I'm behind the curve on record keeping and will be doing something about it this weekend if I can get it done.

I like the false fav angle. Thanks for the the help, believe I can program that in and see where it helps.

To clear one thing up, I can make an odds line....but handicapping this way, not sure number driven enough.

I just finished watching the movie "Contact"(about man's first alien contact) and there's a part of the movie where a character asks a scientist why would a higher life form contact us using mathematics. Answer, it's the only universal language. But, once done recording the mathmatical contact, they couldn't make sense of it. It turned out they weren't looking at the data from the right angle or perspective if you will. I feel that's exactly where I'm at now.
Going off on a tangent.... A horse buddy of mine and I are going to take a look at handicapping the top selections from a reverse view. Approaching the problem from a base of weaknesses in common...not sure yet what we will look at, but the false fave angle will probably be one of them. Should be interesting.

Thank you all. It's been a very interesting discussion. Love to hear more on the subject if any more needs to be said.

Milfredo