Black box software win %

[green80]

I have seen hundreds of adds for black box type software packages but not once have I seen one that posted a win percentage or ROI. I have used several black boxes and I can tell you the win% and roi on each. After I get 200 races logged it usually doesn't change much.

[TonyMLake]

That's odd. We just had an entire thread about how difficult it was to find "black box" sofware, as most handicapping software had many options and many settings - there was quite a learning curve and hardly any black box stuff that just picked for you.

Several people were actually having some difficulty finding black box software, in fact, and most had to settle for normal handicapping software and adjust the default settings with guidance from some of our regulars.

I'm curious what the names of the black box softwares you found and tried were. Would you mind PM'ing me their names? I'd like to check them out as well, as black box handicapping software is an interest of mine.

Thanks

Tony

[GameTheory]

Probably system mill crap...

[green80]

Just use the default settings then it's a black box.

[Flysofree]

The type of software, I call a "black box" only requires the user to decide to bet or pass the race.... It does not require one to figure out which pace lines and how many pace lines and if he should use one setting or the other setting unless it's wet or fast dirt. Look at the odds board and say this is a win bet or I must pass.. That's just the way I see it.. Otherwise just use the Morning Line Handicapper's line...

[jk3521]

It can't be considered a "Black Box"selection unless default settings are used. And that would be for all races unless FTS are involved. Also no odds restrictions.

[raybo]

I have seen hundreds of adds for black box type software packages but not once have I seen one that posted a win percentage or ROI. I have used several black boxes and I can tell you the win% and roi on each. After I get 200 races logged it usually doesn't change much.

Of course, once you have 200 races run it takes much longer for individual race results to substantially affect the overall hit rate and ROI. But, that can be a result of either low results early, within that first 200 races, or high results early within those first 200 races. Much more often than not, I'm confident that those systems you looked at produced negative ROIs over a long run, especially if you included the first couple of weeks of meets, and the last couple of weeks of meets, as those 2 time periods often present some very large payouts, you either got lucky and hit some of those, or you didn't. Chances are, you didn't.

When I test tracks with my black box, unless it is a track that has a very short meet, I ignore the first 2-3 weeks of meets, and I often stop playing just before the end of the meet (because horses are moving in or moving out of the track, along with trainers and jockeys, and weather conditions/surface conditions change dramatically from the very beginning to the very end of the meet), so I look for tracks that test positive, during the "meat" of the meet. Large hits that happen early or late can skew the overall results dramatically, and cause you to make very uninformed decisions on the overall validity of the system/method. Now, if a track produces large hits throughout the meet, that's a different story, that is not a case of outliers falsifying the overall results.

Most black box (system) sellers either report inflated numbers (cherry picking results), or none at all because the results were poor and reporting them would not help them sell the product.

[green80]

I have yet to find a black box that would give me a positive roi, with or without the first or last 2 weeks of the meet. I have not seen any that give you any type of win % or ROI in their advertisements. However I am testing one now that gives me almost 70% winners in the top 3 selections with more than it's share of long price horses. Just have 100 races logged so there is a way to go before I would declare it reliable.

[traynor]

The reason ROI statements are rare (or nonexistent) is that the ROI is based on a specific sample of races, and rarely (if ever) repeats that performance on different samples of races.

With all due respect to the creative "explanations" of why things work differently in horse racing than they do in every other field, the root problem is that most horse race "research" is based on back-fitting to a small sample of races that is unrepresentative of other samples--large or small. Specifically, developers go as far as finding a "positive ROI" in a small sample of races, believe (honestly or otherwise) they have done something wonderful, and declare the work done. And (usually) the wonderful app for sale to anyone with the price.

Statements of ROI would have to contain so many caveats and declarations that past events do not predict the future that no one would buy it anyway, so the warning that an app "is not intended as an inducement to wager and should be used for recreational purposes only" (or something equivalent) is simpler.

The techniques used are not much more sophisticated than Ray Taulbot and many others were using way back when. Sloppy research and approaches that totally ignore the basics of predictive analytics--whether in computer software or pen-and-paper--do little more than provide more funding for the whales with their applications.

[raybo]

I thought that's pretty much what I said. Of course, you did stick the lack of formal educated research stuff in there (predictive analytics - LOL).

[DLigett]

I have seen hundreds of adds for black box type software packages but not once have I seen one that posted a win percentage or ROI.

I posted some win rates and ROIs here last month:
http://wagermathematics.com/blog/top-blog/

...don't expect magic under these circumstances!

[green80]

I posted some win rates and ROIs here last month:
http://wagermathematics.com/blog/top-blog/

...don't expect magic under these circumstances!

I stand corrected.