Nathan,

This would be the main thing I would like to see added to the Simulation Tuning module:

The addition of filters, such as - an odds range to only tune on a certain odds range of winners. I tried doing this by hand but I can't delete races so that I only end up with the races I want. For instance, I wanted to try a simulation on all races won by a horse that won at an odds range of 10.0/1 or higher. It would be very revealing I think.

As a general rule, one of the first things I noticed is something you mentioned a few days ago about using the Sim winner at 10/1 or better. That does work. It works even better if you have a solid block of recently tuned cards for that track, and it doesn't have to be the most recent block of cards either, just a block of 10 - 15 tuned cards used as a control, run the Sims, bet only the first when going off at 10.0/1 or higher. Nice little play and it puts money in my pocket. That makes it real nice.

The only real problem with that play is all the time required stalking the tote board(s) because on the weekend you have to have two or even three toteboards open at the same time.

It is possible to use the SPAT to ferret out where the plays are but what you don't get is an idea of the types of factors that win those races over the long run. But the SPAT will serve to point to the right types of races - and it does it very well. For instance I've always felt that the bias toward fast tracks was unjustified. SPAT told me differently. Same thing with Dirt versus Turf. SPAT said, Dirt only dummy. So, it's only Fast Dirt tracks now - unless it's at the FG.

Another two bits,

Bob

Bob,

I for one would like to do some testing of using the same data without tuning and with tuning for say a 10 day period(or whatever). In otherwords, determine the users that are interested in doing some detailed testing and split up the workload.

For example, two users could work together on one track such as Belmont and one could test with tuning the other without. In a way, the without tuning would be a control group. Say we take the first thirty days of Belmont and tune are data with the first 10 days of that time period. Then the tuning database user would use that 10 day tuning over the next twenty days(or two 10 day periods) and track the success of the tuning using spat. At the same time, the non-tuner would monitor the same 20 day time period(again broke into 10 day periods) and try to confirm if the tuning is better or worse than just using the simulator without tuning.

This same testing could be completed on as many tracks by the same two users or others that want to share information. One problem we might run into is the differences between Bris and Tsn data and agreeing upon what should and should be tested.

Again enjoy your thoughts and hope all of the users are interested in working together,

Bill

Bill,

Sure thing. I'll go along for the ride.

I am just finishing doing some of what you refer to in the background. What I have found so far is that a base, like you mention, of 10 days worth of tuned files works much better than using untuned files for the whole 30 days. But, who knows, we might find something like the "pot 'o gold" at the finish line (it ain't at the end of the rainbow, I've already checked that).

What I have not tested is using untuned periods for testing.

Now there are a lot of things/ways Nathan has set up the tuning that I've changed. For instance, here are the settings I'm using right now:
ROI, Find Most Suitable Race (I like that paceline selector better but testing it both ways might be interesting), no off tracks, and on Pace Bias Tuning I'm using .1, 3, .1 . But if you want to spearhead a full-blown test I'll do it whatever way the group (or us) decide is best.

It would be nice to have several others join in so we can test the heck out of multiple settings - who knows what we might discover? If it increases my ROI .10%, count me in.

So how about it folks, how many of you would like to jump in the water with Bill and me? The water's not deep and it is nice and warm. Should be fun. Since Bill is heading this (you're elected by a one vote margin - sort of like The President) either post here or send e-mail. Smoke signals don't count unless you're full-blooded Native-American.

Thanks for considering it,

Bob

Bill,

Before I forget here, I am using the TSN files.

Bob -

Bob,

I really don't want to be the president, but lets see who else responses. I am using Bris and only have datafiles on six tracks NY, NCal, SCal, Chicago, Florida, and Kentucky circuits. Started downloading beginning of August and keeping database current.

I have free long distance cell minutes every weekend, so once we see our response lets touch base(Sunday night would work best for me since I handicap Saturday night for Sunday).

Okay everyone this is an opportunity to work together!!

Looking forward to it,

Bill

Would love to but never know where I may be from one day to the next. Very good idea. I better pass don't want to screw anything up.

Count me in. I'm using Bris data.

Coal Miner,

I can't speak for Bob, but I don't think we are looking for an every day commitment. Like most people I have an every day job and family that require my time too.

Just seems to me we all want to maximize our profits with this great program and sharing the workload on some of our ideas can benefit all of us.

Anyway the worst thing that could happen is it doesn't work and we are back where we started.

Now how about the rest of you?

Cat got your tongue,

Bill

I will give it a shot. Count me in. I am using TSN Procaps data.
Bill you got two votes for president now! This will take the ability and patience of a school teacher to run.
I will take So. Calif or No. Calif tracks. So far So. Calif is better with untrained settings but have not tried default changing. After ants this should be a breeze to do and maintain.

Ok, I will do what ever. I have no day job that is why I am never home. Like today going to Oklahoma to OTB. I will do Delaware and what ever and I use Bris drf files. Send Info. to ncce1@mchsi.com