Having real pool data helps because the winners true odds were 3.662325382 . allowing NYRA to keep $960 of money bet on the winner.
Paying to 2 decimals as tracks do, NYRA kept the $0.012325382 of $77908.33333 wagered, which is $960, and I show in Tab: Skimming R37
That's possibly where pieces of rebates come from?
I don't need the previous math equations that I used with the mythical data. This data is fine. I'm doing more, but this seemed interesting enough to share.
I have the race chart attached, all the calculations in my excel can be cross referenced from chart data.
The 5, Heavy Meddle was bet awkwardly. That indicates algorithmic betting to me. Also probable 2 or more syndicates bet the horse right at 5.60, cutting his odds in 1/2.
19% of the pool was bet in the last 30 seconds. 36% of Heavy Meddles bets were made in the last 30 seconds. Somebody or somebodies thought this horse was worth a $20,000 bet and that makes the horse interesting to me. I took a deeper look at him
The 5, Heavy Meddle was bet awkwardly. That indicates algorithmic betting to me. Also probable 2 or more syndicates bet the horse right at 5.60, cutting his odds in 1/2.
19% of the pool was bet in the last 30 seconds. 36% of Heavy Meddles bets were made in the last 30 seconds. Somebody or somebodies thought this horse was worth a $20,000 bet and that makes the horse interesting to me. I took a deeper look at him
How do you know that 19% was bet in the last 30 seconds? Or do you mean that 19% reached the tote system in the last 30 seconds? (Is that before the gate opened?)
BTW, I am thrilled that you are getting something out of this.
In 1986 Asch & Quandt did a study that successfully determined that players that wagered in the last 5 minutes were smarter than the earlier bettors.
Of course, today we know that they'd have been better off using the last minute.
I introduce this idea because I think this will point in the direction of truth. (Whatever "truth" is. LOL)
So, first, my sources assure me that the guys I refer to as The Whales now wager all their money late. They didn't used to. They used to feed it in a little at a time, testing the waters as they went.
Now, there are also many Junior Whales (for lack of a better name) that between them make the equivalent of another whale. I assume that they do not bet all their money late but have no way of knowing one way or another.
Permit me to propose a scenario...
First, assume that at a particular track (win pool):
1) 32% of all money is whale money
2) 72% of all money is wagered "late"
(meaning "too late to show before the gate opens)
3) That means that 44% of all money wagered late is whale money.
The questions:
* How good is that 72% compared to the 32% that was in the pool when you were able to actually make a bet?
* If you make bet at the last possible second, when does your money show on the tote board?
Generally if I model something and its interesting I'm using AWS Computing Cloud, pay as you go, al carte. The ML in Amazon Sage is heads above anything home-brewed. The built-in machine learning algorithms allow to train quickly with not a lot of data.
Microsoft's Machine Learning Server and Microsoft R Server also expose's ML algo's that can learn on hyper speed.
How do you know a wager was made in the last 19 seconds?
The data is streamed so its real time to the observer, but no doubt there is lag, or an intended buffered time.
This was a 6 horse field which did provide a good opportunity. "Going in the gate" data was captured, and the 6 horse field loaded smooth and quick, so it was a 30 second clean cut. As opposed to a 10 horse field with loading issues.
From the 30 second streamed pool data time stamp, and with the gate opening, ~19% of the total win pool was added to the tote. Whether it was bet 60 seconds previously and arrived in the stream 60 seconds after bet, I would just need a better pipe, or fire hose hookup.
Overall, the Data's Hygiene is sub-par and cannot be counted on at this time.
I'm note sure what the criteria would be or where to go to for API access which would eliminate the lag *mostly.
If you look at TAB True odds in the excel, its using real dollars. Not odds. I did this in reverse of the previous version. There I did not have pool data so I had to calculate percentages using mythological pool data.
With real tote data, and this is 100% real , it can be run against the equibase pool data. I have 2 columns hidden in the Skimming tab D and E with Place and show pool. They all reconcile to the penny with equibase pool data.
In the true odds tab, you see the money coming, its pre-post time. Real money, not guessing. This is the data that NYRA streamed to me, its real.
The 6, Mighty Zealous was claimed from Gary Contessa. Purse was 90K, claim price was 40. he finished dead last. No purse money and no horse = bad day for Gary, bad meet. 1 for 28, 4%
My theory’s on the wave generated energy and timing data.
Quote:
Originally Posted by Fuss
No takers on the GPS? I thought about it when I met a few Shot Link guys at the Honda Classic a few months ago. I went to watch Tiger's 3rd round.
Advanced GPS or specifically DGPS. (differential GPS) has a horse racing angle.
The common misconception that speed measured via GPS is done so as a function of position against time. DGPS isn't measured like this, it uses Doppler, and as a result, generates 2D course over ground data, vertical velocity, and velocity rate of decline.
The data is measured in wave’s, allowing more precise energy rate of decline. Like measuring the pitch of a sound while its moving away. The data points are endless.
Additionally, GPS uses the atomic clock, race track clocks are not needed. Just common starting point and common end point . (lat/long) 6 F, 7F, 10F , don’t matter.
.
My simple theory’s on wave generated energy and timing data are above.
Holy chit, saw this today! This is where I’m going with my game.
I actually started sketching this out a few months ago. I’m curious as hell what they’re doing. My model uses mostly GPS but also requires some ground-based high frequency lasers. I’d love to learn more and convey my concept model.
Don't want to take this thread off track but just a note regarding The Golden Gate Fields GNSS/GPS data Fuss was referring too which is (im sure) being provided by a UK company - Total Performance Data - who have been supplying punters here in the UK with sectional time and positioning data from 6 tracks (mostly All-weather) for a couple of years now. TPD's website is here http://www.totalperformancedata.com/ and their output is shown on AtTheRaces - i have attached a few screenshots as an example of a 6f race showing the first 4 finishers. New metrics such as Stride length and Cadence and Finishing speed % are catching on and providing some hearty discussion on Social media and changing the way punters here view races. The data itself has had a few teething problems in accuracy , but generally seems bang on the money now. There is a big hope that the other racecourses here will follow with this as it is essential information from nearly every aspect of racing and can even help with stewarding but so far the authorities (BHA) have shown very little interest in promoting this and the majority of tracks including all major ones only have final time data. Timeform UK have went as far as supplying sectional times for this from video and there are some of us who do it privately but camera angles , lack of furlong markers etc provide their own problems.
I think this is on subject. Actually Dave's conclusions are spot on, in that, all the races, and most of the money, is intelligent money.
Thanks for the link and images. I'm in the middle of some research on pools, and the GPS work that I've been doing.
One of the idea's I had, was that we could actually go back and re-time every race ever run. In the same way we can go back and tell what the weather was on a certain day 300 years ago.
And do so more accurately. *except in cases where the gate was mis-placed.
I *think they're already playing with this concept by comparing Secretariat from 1973 to the 2015 American Pharaoh's Belmont Race. Yes, this is cool.
I think this is on subject. Actually Dave's conclusions are spot on, in that, all the races, and most of the money, is intelligent money.
...
One of the idea's I had, was that we could actually go back and re-time every race ever run. In the same way we can go back and tell what the weather was on a certain day 300 years ago.
...
Anyone ever punched a clock for a job? Yes? So we have a historical record of exactly where you were in 1973 when you arrived at work. All I'd need was where you started and I could tell you exactly how long it took you to get to work.
That's it. Just the lat/long of the starting gate and the final time. We could back it up and be dead on.