Trifecta Study on our 5000 Race Database
 

I asked ChatGPT several questions:
Using our database and dividing results into Sprints (5-7 furlongs) and Routes (8-12 furlongs), Turf, Dirt and Poly using the Leverage Pace tab and the results information, determine how often W , P, and Show occur in the top 5 rankings in the following categories, BL/BL, Math (the sum of VDC, CSR, PL and BPP), RX3, Leverage (the sum of VT, CVT, BVT, BVTR), TP (TPR), and ML. Then in summary how often does win appear in the top 3 of at least 3 of the 6 categories? 4 of the 6 categories? Do the same for place and show. Can you do that and if so let’s do it.

Wonderful data! Thanks! Using this data please give me an easy to use report that is sharable with novices to using the program and gives us a strategy for finding the trifecta.

I then asked Chat GPT to replicate the study including Total Energy (TE) as the 7th Category.
Here is the total report.Attachment 56101

Here is a Summary Flow Chart.
Attachment 56100

Attachment 56099


Notes:
1. I have no idea why ChatGPT came up with Rescue horses as a title for horses to consider in the bet.
2. Our Database is sitting at Approximately 5000 races.
3. This strategy can easily be used for Exacta Bets as well.

I like what you are doing. I have been looking at exactas in sprint races excluding Maidens, nw2, nw3, turf an on a fast track only. I look for the four horses I expect to finish first or second. I then determine the the two I expect to win. I now have six potential tickets for which I determine the potential will pays for each combination. I look for overlays with conditional wagers. My database is too small to post results but it looks promising. I keep records for sex and distance with the potential to use track and race class as my data base grows large enough. You have given me some ideas. I hope I have given you some.

Great Work Dick,

Nice presentation.

Keep it up!

Howard believed in boxing your top 4 contenders.
There was a while back, someone that was boxing his top 5 in the last race of the day.

The top 5 box on last race was interesting. He played a 50 cent tri, so $60 per bet.
The last race had a much higher percentage of long shots.
He seemed to make money from what he posted. A lot of the races paid over $100 for a 50 cent bet.

Howard's method was interesting as well. Howard believed a lot of the time, the 2nd most likely to win showed. That counter energy placed. Betting the top two early and top two sustained (or presser) seems to make this method logical as well.

Do you have any real world data? Have you been playing real races with your ideas?

I love this kind of research, but without actual data from implementation, I am not sure how useful it is.

btw, if want more real world ideas, look at mark cramers exacta book.

Dan,
Thanks for your interest and questions.
All of the data is real world data based on real world races in our database. Attached here is the Excel spreadsheet that examines all 7 factors that we are using to pick our plays.
Also here attached is another copy of the Guide attached above that AI wrote as an analysis of the data.

Having just done the research on the collected data base, I don't have anecdotal data on my use of the suggested approach, with how well it worked with percentage of wins and ROI. Based on almost 5000 races this the statistically sound mathematical approach. But if I know me, I will make my individual adjustments along the way that may or may not help me. But this is a statistically sound place to start.

Attachment 56106

Attachment 56107

I am also trying to put together a study of ROI on races in the database that have accurate payoff prices. We know all the races since March 1 of this year have correct payout data. The question remains bet structure, do we use 4 horses in a box, do we box 5 horses, do we key one horse in a key box with 4 others, do we use 2 horses to win and 2 or 3 or 4 others to fill the back holes? There are so many ways to structure the bet that any study has to lay out several structures. Then the question becomes which structure and when.

The interesting thing is the data can also be helpful on other bets like picking horses for Exactas, Pick 3s and Pick 4s.

It is helpful to me to know what the relative strength of the factors I am looking at for the kind of horses I need in my bet.

Just trying to put out some solid data that folks can use to make some decisions.

Dick

OK, as I test out the study, today I made 5 Trifecta bets at a $.50 level. I won 3 of the 5 bets and made a profit of $36.54. I should NOT have made that last bet as it was a Turf, and our study shows that Turf is less reliable to this approach. So basically, I should have been 3/4 with a profit of $54.54.

Here is the read out. Obviously, we need more anecdotal research. And I need more data prior to betting Turf.

Dick

Attachment 56115

Good job Dick, nice ROI today.

A bit of confusion on what I meant by real world, my bad... by real world I mean races that we actually bet on, not just data a computer pulls up.