Is it worth it?

[Capper Al]

I'm rewriting my system from scratch. After five years of being profitable, I lost my way- lost money last year. In the book the 'Mythical Man Month' written by an IBM employee back in the sixties, a statement was made that the first program/system is a throw away. One doesn't really know what they will encounter in coding until they actual code. All the meetings and diagrams and discussions and details never seem to be enough. Projects are always getting reworked. And so is mine. I thought about modifying the successful old app, but I have learned so much over the years by reading and sharing in forums like PA that I opted for the complete rewrite. I'm sure that I'm not the only one here who has done a system dump. Handicapping author, Tom Ainslie, used to joke that a system was created every 2.2 days. Oh well. Just like "off to the next race", we can say off to the next system.

[Augenj]

Although I call my current system, Version 3, I actually have rewritten from scratch about 5 or 6 systems over the years. Computers have gotten faster with more memory and more disk storage allowing for a more thorough analysis. Good luck on your rewrite.

[Capper Al]

If I counted major updates or revisions, the system change count would be in triple digits. That's another reason to just start over from scratch. Only God knows what I have done to the code over the years.

[Dave Schwartz]

Al, I hear you.

The ongoing development in HSH was, for many years, 1,200+ hours of "new development." Last year I did less than 300.

[JJMartin]

If I counted major updates or revisions, the system change count would be in triple digits. That's another reason to just start over from scratch. Only God knows what I have done to the code over the years.
Is this a stand alone app, spreadsheet or something else?

[Capper Al]

Al, I hear you.

The ongoing development in HSH was, for many years, 1,200+ hours of "new development." Last year I did less than 300.

Are you at the point where you hire others to code for you?

I guessed that it would take me 6 months to rewrite it since I could copy in the modules. It's been over a year, and it's only about half way done.

[Capper Al]

Is this a stand alone app, spreadsheet or something else?

Yeah, it's stand times two. I pull down BRIS Multicaps into a C++ program and do some calculations and output to a CSV file. Then I upload the CSV into one of several spreadsheets depending on race type.

Haven't got into the interactive tote-board stuff yet. Not enough time for a one man part time development shop.

[Dave Schwartz]

Are you at the point where you hire others to code for you?

I have tried, but I spend so much time explaining what they need to do that I wind up spending half as much time anyway for a sub-standard job.

The exception are the occasional things I do not do well.


Currently I am in the process of changing our files to an Oracle-type system.

I don't mean we're using Oracle. I mean that I am writing a multi-dimensional database approach myself. The goal is to be able to search and filter the data on "likeness" to something else as opposed to absolute queries.

Nightmare of design.

[Capper Al]

I have tried, but I spend so much time explaining what they need to do that I wind up spending half as much time anyway for a sub-standard job.

The exception are the occasional things I do not do well.


Currently I am in the process of changing our files to an Oracle-type system.

I don't mean we're using Oracle. I mean that I am writing a multi-dimensional database approach myself. The goal is to be able to search and filter the data on "likeness" to something else as opposed to absolute queries.

Nightmare of design.

This sounds like rocket science. During the Dot Com years, I went around the country setting up data warehouses for Fortune 500 companies. My racing data is nowhere near ready enough for something of that scale. Good luck!

[raybo]

My own experience with creating racing programs started with a very simplistic manual entry "Basic" language program written on a Tandy using cassette audio tape recorder. I actually started before that on a TI-99 with cassette tape, but that one went absolutely nowhere. With the Tandy, I did not have the ability to load race data so everything had to be manually entered from the DRF, and the program basically asked questions and I answered them, which resulted in a ranking by grade. Then I discovered Excel and decided to just create the PPs, as close to the DRF layout as possible, so that I could manipulate some of the raw data. Still did not know how to import data though, so everything was still manual entry with some formulas for calculations. I don't remember how many versions of that I went through before ever learning how to use macros to import data files and other things, but that was a huge step forward. That resulted in my first "real" racing program, and that original program went through probably hundreds of minor versions and at least 4 major versions/rewrites, that I remember. That one was named "RaceCap". Then the "AllData PPs" project came along, and I don't know how many versions of that we went through (HCap and I) but it was a bunch! Then finally the "AllData RS Black Box", which built on the "AllData PPs" (renamed "AllData J1") core, I quit numbering versions of that very early and just kept modifying and refining many, many times with slight changes in the workbook name. I probably have 8 or 10 GBs of those old versions on my computer and an external hard drive. I'm always doing things with it, mostly for individual tracks, so it takes a while to run enough races through it to get an idea of validity or not. I use to spend 10-12 hours a day (plus my actual wagering time), 6 or 7 days per week just fixing and modifying/updating workbooks. Now, I sometimes don't do anything for a week or two, so the workload is much lighter. There is still regular maintenance activities to do, but most of that is all automated so I have much more free time than I used to. That's nice!

[DeltaLover]

Currently I am in the process of changing our files to an Oracle-type system.

I don't mean we're using Oracle. I mean that I am writing a multi-dimensional database approach myself. The goal is to be able to search and filter the data on "likeness" to something else as opposed to absolute queries.

Nightmare of design.

:ThmbUp: :ThmbUp:

That's a good approach Dave!

Note that you certainly do not need to create a multi-dimensional database approach yourself! This is a daunting task that will consume all your energy for years to get it correcly (if you will ever be able to complete it!)... Postgresql offers a great Open Source alternative to Oracle, having great support for MDDB: http://www.postgresql.org/docs/9.1/static/cube.html

In my case though, I am not using an RDBMS when it comes to (the very important task) of seeking for "likeness".. Instead, I am using mongodb to store my multi-dimentional objects while building on top of it a matching layer, using scypy and numpy which are providing all the necessary algorithms (like euclidean distance for example) out of the box..

[HUSKER55]

Hey Al,...
before you abandon ship you need to consider that there are a lot of players that are losing right now. I have hit a spell for the last 4 months that I have lost money. I am not in your league but I have lost $1800.

Horses either does not fire or comes in 4th by a neck.

I am taking a break. I was gong to learn python but I found a a guy to teach me who does IT Security for a Fortune50 Company and after an hour of discussion he decided that using SQL and datbase mining would serve me better. According to him it is way more flexible for what I want to do.

It might not be your program. Maybe you should spend some time at the gentlemens club, getting out and about. We are not getting younger. Have you been thrown out of the classiest bar in town yet this year? Have you paid a stripper to sit on your lap and lie her ass off to you yet about what a "great stud older gentlemen make"?

My point is, make damn sure it is your program and not you. Your program worked for 5 years......what the hell happened. Can you define it in very specific terms

Me,...I need to play more tracks and my doctor would prefer I don't watch TV at all. However, he did increase my anti-depressants. :D

All I am saying is to think about it and make sure it is not "cabin fever".

Good Luck.

[Capper Al]

Raybo,

I actually coded the old Radio Shack programmable calculators also. It hasn't been until the last ten years that I decided to stick with developing. The plan now is to finish the rewrite and retire and play with it forever after.

I can't bring myself to pay the big bucks for Allways data. But spreadsheets are the place to end up with the data.

[Capper Al]

Husker55,

The rewrite was actually necessary for where I wanted to take my app. As far as my losing year, it should have been two losing years. My last profitable year I hit two big trifectas saving me, not my style. Also my wager size grew with my new found money, and I moved up from mini-roller to low-roller. This in itself was a big learning experience. I blew it. Instead of increasing my wager size, I wheeled more.

[raybo]

Raybo,

I actually coded the old Radio Shack programmable calculators also. It hasn't been until the last ten years that I decided to stick with developing. The plan now is to finish the rewrite and retire and play with it forever after.

I can't bring myself to pay the big bucks for Allways data. But spreadsheets are the place to end up with the data.

I never mentioned "Allways", my program uses Brisnet single files (.drf or .mcp) or JCapper/HDW data (.jcp) along with the .xrd results files from either source. Personally, I use JCapper data files, the cheapest and one of the best, if not the best, data files available. That was my goal with the "AllData Project", furnish a turnkey file importation method for Brisnet formatted data files, for free, to everyone who wanted it.

[Capper Al]

Delta,

The way I see it, the developer handicapper has one of two ways to start: database or programming. There isn't enough time for a one man programming shop to do them both justice. After both methods( app with database) have been satisfied then one can move into datastores, data warehousing, and data mining. Dave seems to have lots of data behind him, an enviable position to be in. Glad to see someone make it that far. Congrats Dave!

[DeltaLover]

Delta,

The way I see it, the developer handicapper has one of two ways to start: database or programming. There isn't enough time for a one man programming shop to do them both justice. After both methods( app with database) have been satisfied then one can move into datastores, data warehousing, and data mining. Dave seems to have lots of data behind him, an enviable position to be in. Glad to see someone make it that far. Congrats Dave!

Not necessary. The two ways you are referring to, complement each other and are both needed for any kind of modern software development process... I do not think that you need a data-warehouse for horse handicapping needs, you just need to use the proper combo of back end platforms (like and combo of mysql, sqllite, mongo, csv or whatever else) and develop your front ends using some MVC framework (like angular, backbone, knockout or whatever else)... For datamining and "scientic" programming there is no need to reinvent the wheel, as today we have quite a few very powerful open source toolkits that can solve pretty much any problem you can dream of..

[Capper Al]

Delta,

I didn't say it was either or. Both apps and databases serve the capper well. The problem is the time it takes to do them both correctly, or as I said do them justice.