See if I have this right. The deT is using AI to start-- determine the Best Horse(s). Then to determine if the Best Horse is a BALO or not we're using AI's four (4) Green Pace Objects. So at this stage in the process deT is using "pace" versus other factors? So is the deT different than what HSH was doing above? — Jim Pommier

That answer was meant for Saratoga, who does not have anything within several magnitudes of DET to answer his questions.

Please forgive if it sounds like I am nitpicking.
You might have the meaning right, but I need to be sure, lest you or others are confused by my lack of a straightforward answer.

1) The BALO flip tells you when there is a... well, a BALO or not.
2) The AI Best Horse flip gives you HIT RATES and/or VALUE.
3) However, the AI Best Horse also serves a distinct purpose within the BALO, but only as it pertains to the BALO candidates.

The BALO PROCESS includes:
1. AI LINE Magnitude=1 to select the BALO Candidates.
2. Then, the AI BH Mag is used to test the BALO Candidates.
3. Finally, the 4 Pace Objects are also used to test the BALO Candidates.
(As we've discussed, there are some rule changes for E and EP objects on turf.)





_______________________
Has this response improved your knowledge?
(Or possibly made it worse?)

The issue is really SESSION PLAY, which all of those support nicely.

As I alluded to, I use something else.
It was built into HSH and will see the light of day in DET pretty quickly.

Probably not on June 28, but probably the week after.

Did I get the point across in the video?

If this answer is yes, I expect that it will be vitally important that we all work together in order to maximize the power of the de Terminator. I can imagine that it will be difficult and time consuming to discover new angles but if we as a group can mesh different angles together it could be quite powerful, and it would be a pretty fun handicapping journey. — Dave R

This should truly be a very small challenge.

However, I'm sure that COLLABORATION will be HUGE for almost all of you.

Remember... TWO-HUNDRED users maximum.
________________
What will be important - and I've mentioned it several times in videos - is that you do not dump what you're doing now.

What you need is:
1) A Core method of handicapping.
2) A Handicapping Flip that adds new and powerful information.
3) A second Handicapping Flip that adds new and powerful information.

The key to making it work is that 1,2 and 3 should not have a high correlation between them.

IOW, you can't do something like:
1. Best speed rating in 90 days
2. Best Speed rating ever.
3. Best speed rating at today's surface/distance.

While this is an overly simplistic example, it illustrates the point. During a conversation decades ago with the very smart Dick Mitchell, he once referred to this as "FACTOR OVERLAP."

Now my question to Dave is how many more of these types of angles will you be showing us through this onboarding process? — Dave R

Back to the original question. LOL

I spent about an hour while I was waiting for a doctor appointment the other day dictating a list of creative Handicapping Flips, which is my word for them.

I came up with about FORTY of them!
________________
There is a physics principle at play here that I derived from the work of French Astronomer Pierre-Simon Laplace that would indicate you should use an absolute maximum of THREE.

The concept I derived from Laplace's work (1770s on celestial mechanics and the stability of the solar system) allowed me a rudimentary understanding of how many such FLIPS can be applied.

The number I came up with is THREE, and going beyond the second one does not offer near as much gain as the first two.
________________
Lest I begin to reinforce my PA naysayer's opinion, I am only an amateur physics guy in most arenas. Heck, I struggle to read the equations.

However, I am really good at visualizing the concepts and that's where the real payoff is found.
________________
BTW, I the AI that I write is always a Genetic Algorithm. But they are a far cry from the GAs that you will read about in books. (Were you to actually read a book about GAs. LOL)

Now my question to Dave is how many more of these types of angles will you be showing us through this onboarding process? — Dave R

First, I must take issue with the word "angles."
That implies a set of rules that include track/surface/distance/etc. components.

These are concepts that can, with only minor variations in application, be applied across all races.

I understand that this idea of one-size-fits-all is very foreign to horseplayers because we are used to having to make sure that proper context is applied.
___________
"PROPER CONTEXT" is precisely what the AI does best.
In fact, this is the one thing that allows us to not only successfully challenge the whales but to BEAT THEM!

What we're doing is RIGGING THE GAME IN OUR FAVOR.
___________
How can I be so sure of this?

A number of times you've heard me refer to how the Whales handicap HORSES. That is, the HORSE is the entity in their database. Think of it sort of like saying, each factor has weights that are applied to each horse.

Basically, that means one large, global system.
Emphasis: ONE global system.

deTERMINATOR AI creates and manages MANY situational systems.
When I say, "Many" I mean HUNDREDS.

It's engine essentially DECIDES that this is a "type 143 race" or a "type 12" or whatever.

IOW, the CONTEXT is already applied!
___________
The Whales will eventually have to adapt by using AI themselves, but not today

But remember, that is not a trivial undertaking for them.

Think of it this way - they have a winning proposition NOW.
We're just going to be 200 guys.
That's TWO-HUNDRED PLAYERS who will be able to COMPETE with them.

Maybe 30 or 40 of you will go on to play on a professional level. Let's say that it's 40 guys, generating on average $250k per year. That's $10m profit.

Collectively, we do not even amount to a SINGLE NEW WHALE TEAM!.

Imagine that you have a thriving business - making 10s of millions per year. You will be very hesitant to just junk what the most important part of what you've been doing.
___________
Logically, they will eventually embrace the AI world.

But guess how they will logically do that.
They will integrate an existing package... like TensorFlow or similar.

Here's the thing... MY ENGINE IS BETTER.

What?
Well, why wouldn't it be?
It was DESIGNED to do one thing... UNDERSTAND and PROFIT FROM horse races.

There's a guy on PaceAdvantage who has often told me to just give up pretending that I am capable of stuff like this. LOL

Trust me. I am capable.

<More to come...>

The race was CHAOS 5 most likely because there were too many horses in the AI LINE's top 88%.
There can be other reasons but that is most common.
(I seeded that reason in the AI, but it has the ability to modify it.)

Ok watched the next video and seen the button that called Chaos 5. I missed it when you hit it on that Aqu 1. Was the reasoning for them being Balo because it was a chaos race ? — Ken Fee

No.
They are stand alone Handicapping Flips.
IOW, a race cannot be a BALO race because it is a CHAOS race or vice-versa.

I do not recall the race from memory. What was the CHAOS number?
0=No CHAOS
1 to 3 = Specialized CHAOS
4= CHAOS where the winner SHOULD be handicapped.
5= CHAOS where the advantage is in NOT HANDICAPPING.

@Chuck

If you have an interest in getting weight loss or getting healthy, watch this 3-minute video I recorded.
Click here to watch

They're open to the public.
https://www.community.horsestreet.com/discussion/520/determinator-video-list

First, I was watching the videos and understanding the concepts etc. I wanted to have Dave show us or me a picture of what the software looks like in terms of presentation i.e. On your desktop, how it opens, and what the screens look like.. Remember as much as the guts are important, visuals in presentation help the user understand even though it may be complex in nature. i.e. under the hood, but everyone wants to see what the car looks like as far as interior and shinyness. Also, for those who may or may not know. — @Chuck

I'm confused.
Almost all of the videos are live looks at the screens and operations.

Perhaps you mean you want to see all of the operations from start to finish, including downloading and importing race data, etc.

If so, that will have to wait for the actual onboarding process.

I just have too much on my plate right now.

I know had mentioned to myself in teachings and in conversations that he lost a lot of weight which is such a great accomplishment and I was hoping that he could share for those of us if he so chooses, how he did it, his choosing of eating habits etc. Thanks again and look forward to future videos and hopeful some great screen shots of the questions asked. — Chuck

Oh man.
It's been quite the journey.

I'll find some time later tonight to answer this as it deserves to be answer.

Dave

Another excellent question.

Obviously, it has to learn from somewhere, so, yes, there IS a database. It's just not part of the distributed system.

The training module -which is not part of the distributed system either - DOES contain a database.

I actually got the idea from the LLMs - Large Language models.
When I saw they were all pre-trained, I thought of how one of the big handicaps with HSH was that it just didn't come with enough data on day one to study or build much of anything.

REALLY GREAT QUESTION!
The results across all poly is inconclusive, with a tiny leaning towards only a very slight impact change.
IOW, from a modeling POV, there are no hard-and-fast rules that fit, as with turf. More so in turf routes and almost none in turf sprints. (Anecdotally, they are too short and the turns too tight to be anything but a positional battle to the 1st call.)

Instead, the internal "modeling" seems to handle this. Frankly, I am shocked that it works so well, considering that there is no data; no model being kept.

Still, it is capable of looking at a race and saying, "This is what an E-style (or EP, etc) Winner's profile should look like."

But I believe you're real interest is in THE GREEN NUMBERS.
During the BALO Test video, I went aside looked at a couple of dozen turf races (from other days), and the results were really quite obvious - that strict enforcement of the 1s all the way across would be folly.

Remember that the REAL WORK has been done.
That is, the creation of the pace objects and the 1 thru 4 comparison to the whale betting. (Green Numbers)

But personal interpretation of the output is what prevents us all from being on the same horses!
Logically, you will make your own distinctions in your early experiences.


._________________
There is a typo (apparently) in the new pars for WO dirt - which has been corrected.

What are the average winning odds for chaos 4 and 5 races? Also, we know that 88% winners come from 8/1 or better horses in all races. Is it still 88% in all chaos 4 and 5 races? — Dave R

I don't have data with the AI yet, but I can share what I know from my history over the years.

First, a tiny correction - 88% of winners are BELOW 8/1.

Is it still 88% in all chaos 4 and 5 races?
No. It is MUCH lower, meaning that the chances of a longshot payoff jumps through the roof. My guess would be something like 30%-40% or more will pay 8/1 or above.

If memory serves the AVERAGE PRICE in Chaos 5 is something like $28.

You just gave me an idea for a type of workshop we could schedule regularly: STATISTICAL QUESTIONS.

This would be where deTerminator users post their questions and we work thru them live using HSH as the primary tool.

Dave, will these meetings be recorded so that we may go back and review them at later dates? — Dave

Absolutely.
They will always be added to the list of videos.

1. Click inbox on the menu bar.
2. Click start new conversation in the upper left corner.

We need at least one post here in order to send a PM. This is mine. — Red Knave

Discovered that last night.
I'll get Sherri to turn that off.

Curious as to what AI software package you used for your new product? — Slim

It is 100% original code.
Nobody else's libraries whatsoever.

Send me a private message with your info, please.
Click Inbox.

Some of the BALOs get bet way up - 5/1 or higher.

Are these higher prices visible at 1-2 minutes to post
or are the severe odds increases only seen after the gates open?

I bet through Bovada and betting is cut off at 0 minutes to post. If the
horses circle behind the gate for 5 minutes, my ability to wager late
is zero.

Rephrased - will the public show these horses as cold on the board
before the ADW money is "not" sent in? Does the public money always/
usually agree 1-2 minutes before the start with the ADW money? — ponyplayer

I have no idea.
Well, I have some idea.

Please... I BEG YOU...
If you insist on using the GATE ODDS or anything from the tote board... you will prevent this from working AS IT WAS DESIGNED TO WORK.

The idea is that the AI is predicting who SHOULD or SHOULDN'T be bet.

Try to wrap your mind around the fact that this is DIFFERENT than who WILL or WON'T actually be bet.
_______________________
Every experiment I have ever done with BALOs of my own creation failed miserably when I used gate odds. I KNOW it seems logical that the tote would improve things, but I PROMISE it will not.

Remember that the AI is actually using my approach to this, except it had the ability to change the numbers and modify the Ranks & Mags.