So Now I Have a Database

[NormanTD]

Okay, so now I have a database. I can find out how the fastest horse with a trainer win percentage of 18% or better who is coming back in 37 days (the horse, not the trainer) did at CDX or TAM or xyz... fast track, second best Quirin speed number and gained 3 lengths in the previous race...and on and on and on.....

But we all know those "filters" don't work because it's all conventional thinking and everyone can see those items. I unfortunately find that my thinking is pretty inside the box and conventional too.

So my question is (are?):

How best to use a database?

Maybe the more extreme question is how to think outside the box (how to NOT think like the crowd?)?

Or maybe even, "What questions should I be asking here?".

I would appreciate most any comments and/or suggestions.

Thanks in advance!

[Dave Schwartz]

First, so we can understand the scope, how much data do you have?

[Tom]

Read the Bet This columns in DRF for ideas.

Look at individual trainers.
How does Joe Trainer do with 3 yos in dirt sprints, after 120 days off?
How does he do turf routes to a dirt turn back in 45 days or less?
Look for reasons to throw out short priced horses as well as bet long shots.

Think specific rather than generalities.
Most people don't have that stuff.

[Poindexter]

I do not think there is a standard set of questions to ask. I think the ideas will come up as time goes on. You see a e type runner who usually fades at 6f stretch out at 7f and of course fades again and then drop back to 6 f and all of a sudden instead of fading, he actually rates a little off the lead and rolls by late to win. Maybe we have something. Time to look at the database. See a horse fire a big shot off of a layoff and then comes back with 2 solid works, will he bounce(time to check the database). S type runner stretches out to a mile for the first time. Might want to check the database...........There are so many things that come up on a daily basis, you will want to play around with. I have never used a database for racing, so I have no clue what you will find but every time a horse wins that sort of surprises you, it is time to see if there is something you can learn from it. I think a lot of stuff will leave you shaking your head thinking "why am I wasting my time on this", while you will find a few nuggets that will make the effort well worth your time.

The problem with a racing database vs say a sports betting database is that in sports betting you have a line that basically equals everything out to pretty close to 50/50 propositions. So if you find that NBA teams playing back to back off a loss(purely hypothetical) only cover 45% of the time over a large sample, you found something special. With racing, a horse could be 13-1 but really should be 30-1, or a horse could be 2/5 and really should be around 4/5. Thus your roi numbers can be very deceptive. Also with sport betting a team covers(wins) or doesn't cover(loses). In racing a horse could run 2nd by a nose, or run unbelievable in defeat dueling through ridiculous fractions only to be nailed late by 2 closers. In your database it will show up as a loss, even though the horse ran amazing. You might want to look at roi figures not only for win, but for place to take care of some of these types of situations. In sports if a team plays amazing they likely will cover the spread.

[acorn54]

i will add my two cents. i am not one to "open up" about how to tackle the warfare in the parimutuel markets. make no mistake that is what it is. you against every other person betting into the parimutuel pools.
we had this discussion in the jcapper forum, jcapper is a database driven software. jeff platt was generous enough in the "private forum" to give out a profitable database model of the past 3 years consisting of many thousands of plays. sure enough within a matter of a month or two the model which had been showing a profit for the past three years went downhill roi -wise.
the win percentage stayed the same.
this also happened in mid -2005 with the "j factor" in jeff's software program.
jeff was principled enough to stop the sale of the software with the "j rating" to the original 100 owners of the software.
what i just wrote i hope stays with you during your very long work in finding anything of value in your database research pursuits.

[raybo]

If you don't have specific questions about races and racing, then you don't need a database. If you do have specific questions, then a well designed database is exactly what you need to answer them. It's really that simple.

In other words, if it were me and I was considering creating a real racing database, I would be thinking about, and writing down, what I wanted to know, first, before I created the database. Otherwise, I wouldn't spend the time and suffer the inevitable frustration involved in creating a racing database, in the first place.

The database that I have in my track workbooks was designed to capture specific field dynamic data, to be used for eliminations from win contention, and that is all it is used for, and that meant that I didn't need a full blown database app, just a "list" in Excel, linked to a pivot table. All of my "what ifs" are formulated in Excel itself and manipulated into a potential rankings method, then "tested" against many cards and results, and the number of plays, hit rates, ROIs, and profit/loss are outputted to a record keeping sheet within the workbook. If the results are positive they become a playable rankings method, if not they are discarded.

I have 11 playable rankings methods, all testable for each track all in the same run through the batch of cards and results files, one of which is a "variable" method that I have 6 other optional rankings methods tied to, and I also use that variable method for testing new ideas.

[NormanTD]

First, so we can understand the scope, how much data do you have?

2014 BRIS files and the exotic results files for most tracks and the same for 2015 so far.

[NormanTD]

If you don't have specific questions about races and racing, then you don't need a database. If you do have specific questions, then a well designed database is exactly what you need to answer them. It's really that simple.

In other words, if it were me and I was considering creating a real racing database, I would be thinking about, and writing down, what I wanted to know, first, before I created the database. Otherwise, I wouldn't spend the time and suffer the inevitable frustration involved in creating a racing database, in the first place.



Thanks Raybo. The database was as much an intellectual pursuit as it was (is) a handicapping tool. Sort of, "I wonder if I can do this?" and I think I sorta did. I call it the Petroleum Database because it's crude and unrefined.

One of my intents was to test misc handicapping methods and/or spot plays that I've purchased over the years and of course, they were not worth the paper they were written on. So I wanted to get some professional help and see what else I could do.

[JJMartin]

2014 BRIS files and the exotic results files for most tracks and the same for 2015 so far.
What program are you using with your DB?