Joe Zambuto program

[redeye007]

Originally Posted by redeye007

Here's a better way to self tune for anyone interested in trying it. you'll need a decent size database of race results. parsing the free results is a good way to do it since the purse value is there. focusing on the target race that we want to create a new weight for perform the search in self tune using sprint or route turf or dirt and plus and minus 20 percent of the target race purse value. for example the target race is a 6 1/2f race on the fast main track with a puse of 50000. I set up the search like this: Dirt=fast=<=1540 =40000-60000. I like to see a minimum of ten races in the search parameters. now I do q-check then f-tune, name and save. now I have a good weighting criteria to handicap this race.

Any JZ softare users that would like to test the criterias using the 20 percent purse method that picked the pick 6 winners 1st or 2nd at Del Mar on 8/15 pm me.

[redeye007]

Self tuning using the 20 percent purse method seems to be picking more winners

[redeye007]

Here's an example of the power of the workout rating in the program. DA Boomer in the 7th had the best workout rating and was a first time starter and returned $28.40 to win and keyed other nice exotic payoffs.

http://www.paceadvantage.com/forum/s...ad.php?t=74422

[Partsnut]

redeye007:

Here's an example of the power of the workout rating in the program. DA Boomer in the 7th had the best workout rating and was a first time starter and returned $28.40 to win and keyed other nice exotic payoffs.

Here's another example from one track betting all 9 races and hitting five trifectas. Believe it or not. 08/24/10


http://www.box.net/shared/j5tpre73ho

[Partsnut]

Here's a trifecta I just hit in the 4th at DMR. This one race put me on the plus side for the rest of the night.

2,4,5 paid $218.80

My print out is available for those that want to see it - Just drop me an email.

[Partsnut]

This race should get your attention.

I had the complete order of finish for the first 4 in order.
I didn't play the super but I did play the tri and caught it twice.

[JimG]

Originally Posted by Partsnut

This race should get your attention.

I had the complete order of finish for the first 4 in order.
I didn't play the super but I did play the tri and caught it twice.

Congrats on the hit(s). Are you using Zambuto out of the box or are you doing some tuning?

Jim

[redeye007]

Nice going Partsnut

[Partsnut]

Jim G:

Congrats on the hit(s). Are you using Zambuto out of the box or are you doing some tuning?

Jim

Jim, you need to compile a lot of races for an individual track for the database to accurately predict each distance and surface. I took a shorter route and played around with the criteria files which governs the databases predictability and custom expressions that I felt would be meaningful. The criteria files can be adjusted to what you think will be effective. If you put a lot of time in and play around with the numbers, eventually you will be successful. I've tuned a couple of tracks. Keep in mind that certain software products are good for win betting and others for the exotics. My success was due to extensive testing, backfitting and trial and error.

[Partsnut]

DMR-Race (8)

3 tri's out of 8 races.

[redeye007]

I like to play pick 4's. sometimetimes all it takes is using the top 2.

[redeye007]

Here's a tip I learned for self tuning if you are using the BRIS exotic results. the exotic results do not include race type or purse value and these are two of the factors you want to self tune with. not a problem. just open the database manager in the results section, locate the race card after the exotic results have been added and just type in the race type and purse by placing the mouse cursor in the appropriate cell. now this info will appear in the self tune module. Parsing the free results does include this pertinent data though. I f you don't like the text in a cell like "turf(rail at 20 feet)" you can change it to just "turf". whatever you type will appear in self tune. I'm presently experimenting with purse ranges again. Using the exact purse value you can set the range for self tuning, for example (15000-22000) and all the races with purses in that range will be listed. this may be a better way still. I'll post back the results of my testing. tell us your best way of self tuning.

[redeye007]

Here's still another way to try to self tune. open the self tune module. suppose you want to tune for a target race of 7f(1540), dirt, fast, 20000-30000, claiming. set up these parameters for race segmentation. now click set factors and set all factors to -1. next select all races and click F-Tune several times until the win percentages don't change. then click set factors again and change all the -1's to zeros. name and save if the win percentages are higher. this method seems to yield slightly higher win percentages when tested against the database.

[redeye007]

Here's more interesting data regarding self tune. I segented my delmar database for all 5 furlong turf races. the win percent was 46.67%. I then ran Q-check then F-tune with all factors set to zero and increased the win percentages to 53.33% with the new weighting criteria. I then set all my factors to -1 and ran F-tune and increased my win percentages to 73.33%. then I changed all the -1's to zeros and ran F-tune again and the win percentages stayed at 73.33%. In other testing the win percentages actually went higher in some cases using this method to self tune twice.

[Partsnut]

Before I actually use a piece of software I will try and learn every facet of what the softwares capabilities are and what it can do.
I've done just that with the Zambuto handicapper. I have it about a month and
I can say, without hesitation, that this is a very flexible tool.
Today, I wanted to see how value bets would work out so I made a few adjustments to my criteria files and built a new custom expression and filtered each individual race as I saw fit.
Please keep in mind that I have not built a database yet for any particular track.
So I'm working off the basics that was supplied with the program.
Here are the results from the projections I posted on my site, before the fact.