Should I Be Content?

[jackad]

I've been tracking a software program over the last 30 days. Being very selective about which of its top picks I bet, I show a win frequency of over 30% and an ROI of over 30%. The drawback is that handicapping an average of about 10 tracks/day and being this selective, I only come up with 2 bets/day.

How would you evaluate such results? Thanks.
Jack

[Dave Schwartz]

Jack,

I'd say you are about to be a wealthy man. (I am not joking.)

With a 30% hit rate and a 30% ROI and 2 plays per track... Let's play "just" 8 tracks per day and make 16 bets per day, 80 bets per week.

With a +30% ROI and odds of 3.3:1, your optimum bet is 9% of bank. Let's bet less than half of that... 4%.

Let's consider a 10-race segment of 3 wins, 7 losses:

Losses = -4% so you wind up with .96 of BR.
Wins = 4% x 3.3:1 = +13.2% = 1.132

So, a 10-race segment gets you:

(.96 ^ 7) x (1.132 ^ 3) = 1.09 growth factor per 10 races.

You get to do 8 of these per week.

1.09 ^ 8 = 1.99

In other words, your bankroll doubles once per week until you max out. Assuming a starting BR of $200 and an ultimate max bet of $200 (your bankroll maxes out at $5,000):

1. It will take you less than 5 weeks to escalate your BR to $5,000.

2. You will then have a nice income of $4,800 per week.

Of course, with average odds of 3:1 you could certainly bet MUCH more than $200 without hurting your percentage much, but $240k per year ain't bad.

Congratulations.


Regards,
Dave Schwartz



Regards,
Dave Schwartz

[jackad]

Dave,
That was two bets/day; not two bets/track. Sorry if I wasn't clear about that.
Now what do you think?
Jack

[VetScratch]

Jackad,

What you seem to have adopted is a personal approach that has tested profitably so far. If your results hold up over a SUSTAINED period, your satisfaction should measured in terms of recreational and financial value.

Being very selective with 10 racecards a day may fail to provide sufficient recreational value to hold your dedicated interest.

However, if you can sustain a 30% ROI without undue limitations on your wager units, 30% daily interest on your investment beats most other alternatives.

Because you have to work hard, you may find that your approach will have lasting value since you will have few imitators.

However, if you find the drudgery unbearable, just don't give back what you have earned by going on a wild recreational spree!

[Dave Schwartz]

Jack,

You mean you look at 90-100 races to find two plays? TWO plays?

You, my friend, will have to become the patron saint of patience. (I'm not Catholic... Is that saint taken already?)

I will still offer my congratulations, but now I have to add: "... that is a lot of work for $30k per year."


Regards,
Dave Schwartz

[GameTheory]

IF those two plays are easy to identify (i.e. without watching the odds for all races all day long), then I would be very happy. Bet them, and be happy. That's two solid plays a day you've got in your pocket.

Now try to come up with some more strategies with similar success rates. Incrementally build up a "stable" of spot plays like this until you've got 15-20 a day. I took this approach for years and it worked quite well for me (of course it can take years to uncover good plays as well). I'd just have a checklist of plays -- all of them rare, but enough separate angles that it really adds up to something.

[jackad]

Thanks guys (and gal) for your useful comments. Food for thought.
Jack

[Hosshead]

Jack, I think GT is right on. Be happy with your 2 plays a day,
and work from there. Just curious, how much time/trouble is it
to find those 2 plays? Also, when collecting spot plays like that,
it takes a much longer time to find enough plays to get an
accurate account of whether it Really works in the long run.
Also, does your method come up with longshots or favorites?
If there are some longer price horses, be careful,- your ROI may
look good, but : If you happen to: Go on vac. - Get sick-
Over sleep- Have a hangover,- or a hundred other things, and
you miss a play, and it happens to be a longshot winner,- it will
take a long time to make that up. -- Hoss

[Gekish]

Jack,
Let me add my 2 cents worth. Let me ask a question. Is this back fitted system on 30 days data or a system based on previous data, that is being tested now ?
I have a system that has 86 playable races out 2230 in the last two months. I have divided my races in 2 groups and developed the rules on Group 1 and tested on Group 2

code:
Plays Winners Won Profit
43 16 $126 $40.2 Group II
$86 37% 47%

43 20 $141 $54.8 Group I
$86 47% 64%

86 36 $267 $95.0 All
$172 42% 55%

As you see the win and profit percentage drops for Group 2 and the last 2 weeks it is breaking even. I will test for an other month (1000 races) since with the small number of bettable races I am afraid that my method is result of back fitting.
If you are a recreational player even the Group 2 results are fine, but if you try to generate income it is hard to play for 2 weeks and breaking even.

[jackad]

Hoss,
Ten tracks -- all told about 2 hours per day to come up with my two bets. No watching track odds; I use ML odds as part of my selection method.
No really big longshots; very occasional 10/1-12/1 payoffs.
Jack

[jackad]

Gekish,
Out of 4+ weeks of testing, only one week when this method broke even (no losing weeks). I have no idea whether or not results will hold up over the longer term.
Jack

[Dave Schwartz]

Jack,

A little caution is in order here... If you are saying 30 days of play and 2 plays per day, that's 60 plays.

This is a little premature to be counting the money.

I'd say you should have about 3-5 times that many plays before you will be "sure." That does not mean you can't bet them in the meantime.

I have had systems that produced medium-to-high hit rates and high ROIs at the 500-race mark only to have them die by 1,000 races.

Proceed cautiously, but DO proceed.


Regards,
Dave Schwartz

[Gekish]

Jack,
Are you capping manually from the Form or downloaded data with a database or program ?

[Gekish]

Dave,
As I said before at length, been at this for 50 years. A serious question. I like to have your opinion. Why do systems, angles and computer generated picks suddenly die, even when all the rules or logic make sense ?
This goes for high win % low pay and low win % high pay stuff. I has months of succes with systems just to see them fizzle out. Same track or same circuit same type of races.

Puzzled in LalaLand

[Gekish]

Jack,
Sorry just realized you are using a software program.

Senility is setting in..........

[Secretariat]

Gekish,

I generally try to avoid these things, but your question is a good one. Often we think these things die out because (a) they weren't researched sufficently to begin with; or (b) the hyperbole of the angle didn't match its reality; or (c) it was uniquely designed around a select group of tracks to work; or (d) bettors adapt.

Personally, I lean mostly to the (d) above. A perfect example are speed figures. They are probably the best they've ever been but the common availabilty of these figures tend to reduce mutuels. I also think there is a lot of "accepted annecdotal claims" and a lot of "older research." Doc Sartin once said to me challenge the so-called experts assumptions, and Mark Cramer relates often "find out what the public doesn't do."

My own advice is to look at what most people accept as gospel and explore thopse weaknesses. I'll reveal some research I have been doing now, albeit a little reluctantly. I do so because the occurrence is infrequent.

From the days of Ainslie everyone "knows" it is absurd" to play fillies or mares versus males. It's a toss out, an elimination according to most handicapping gospel. I just ran over 1000 race CARDS, not races, and found that piece of advice has impacted on today's mutuels.

My results playing ALL fillies or mares going into male races was as follows using races from 2001-2002.

87 plays
12 wins
13.79% Win%
176.40 won
1.01 or a 1% profit playing them ALL

Looking at just Routes it looks like it is still applicable to eliminate:

Dirt Routes
14 plays
1 win
7.14% Win
21.20 won
0.76 ROI (24% loss)

Turf Routes
32 plays
2 wins
6.25% win
31.80 won
0.50 ROI (50% loss)

Dirt Sprints
32 plays
8 wins
25.00 % win
82.60
1.29 ROI (29% profit)

Turf sprints
7 plays
1 win
14.29% win
40.80
2.91 (191% profit)

It is hard to get a large database on this since it does not occur that often, but over a 1000 racecards reveals something. My hypothesis is that this auto elimination is becoming a potential new angle for sprints. If this info became widely curculated then I think the mutuels would correct themselves, and the figures would correct themselves. Already since Ziemba's book on the underbetting of the favorite that bias has seemed to reverse itself as well.

Overall, my advice is look at what people say to disregard and DO your own research with recent data.

[Dave Schwartz]

Gekish,

I don't really have an answer to that. I have a few theories but nothing to back them up.

Dave

[GameTheory]

I agree with Secretariat in general, but would also like to throw in the idea that some systems will only work well during certain parts of the year (or possibly the meet). This has to do with the age of the horses and the changing race conditions; the intents & strategies of the trainers, who have different motivations at different times; development patterns of young horses, etc.

I also happen to think (as Secretariat suggests) that the way the public bets is constantly changing over time, but I think it changes on two tracks -- a long-term "X (e.g. speed figures or jockey stats or whatever) is becoming more popular over time" kind of thing, as well as short-term seasonal cycles that occur because the people are always trying to correct what they screwed up in the last race they bet, but since the reasons horses win races also have seasonal cycles (I believe) they are also one step behind.

Certain styles of play will cut a path through all of this and lead straight to the cashier's window 365 days a year. (e.g. trainer handicapping that considers the angles in relation to the current meet, season, & the trainer's stock.) But most need to surf the wave and be adaptable to changing conditions or they will ultimately fail, or only be successful at certain times of the year. All of this can be highly track & class dependent as well. There are lots of variables, no doubt!

Like Dave, I don't have much to offer in the way of backup for these theories (partly because these ideas are not totally proven, and partly because I'm too lazy to provide evidence), but I'll throw them out there anyway...

[Myhorse1_X]

Jackad:

You never did say what software you were using. If you don't want to name it on this message board, I would appreciate it if you would send me a private message.

I have my own software, and I am always checking out other software and compare it with my own.

MyHorse1

[jackad]

MyHorse,
The software is ThoroBrain 5. But I use it in a special way that I choose not to reveal.
Jack

[Topcat]

<That was two bets/day; not two bets/track. Sorry if I wasn't clear about that.
Now what do you think?>

Jack,

If I may butt in here-if you have only tracked this for 30 days that means you have only 60 bets? Right? And you won 20.

I think your sample is way too small to conclude anything other
than what could be a lucky 30 day period. I think it may take you a couple of years of data before you could conclude anything with just 2 plays a day.

BTW: The Catholic patron (AKA model) Saint for patience is St. Monica, who prayed and waited for decades of the conversion of her husband and son. He son who was quite a rogue at the time went on to become St. Augustine.

TC

[Dave Schwartz]

TC,

Yes, I recall that story now.

Thank you.


Dave

[B. Comin']

Originally posted by jackad
MyHorse,
The software is ThoroBrain 5. But I use it in a special way that I choose not to reveal.
Jack

J,
I can relate to your sentiment about how you choose not to reveal how you use TB5.

I use The Sharper Handicapper and Handicapping Magician in combination (even tho' Pizzola says not to use HM in combo with anything else) to play exacta boxes, with an odds requirement on my key horse. I may only have one, maybe two plays if lucky, on a given card, but can do three tracks easily in two hours.

[pmd62ndst]

Sorry to chime in so late on this subject but here's my take:

If you have your database run random queries 24 hours a day, it will come up with something like a 500% ROI for horses with an even number post position below 7, a Last Beyer that can be divided into the day of the month, a jockey that wears red silks on weekdays and a trainer who's last name starts with the letter "L".

You could probably conclude that this magical ROI will not last over time. Any numbers that you get out of computer should be studied and not to be interpreted as "law".

I don't use computers to pick winners, but rather to validate angles and theories. From experience, 60 entries is just not enough.

PMD