Who's a better handicapper/bet maker, HUMAN or PC? - Page 11 - Horse Racing Forum - PaceAdvantage.Com

raybo · 08-23-2015, 05:59 PM

Quote:

Originally Posted by Speed Figure

Why not play Saratoga and Del Mar next week? That would be about 22 to 24 races! that's more then enough to see who comes out on top!

That would be very little better than playing a single race. 22 to 24 races doesn't even account for a drop in the bucket regarding the long term. Variance alone, one large payout, one big loss, etc., could determine the winner of that small sample.

raybo · 08-23-2015, 06:03 PM

Quote:

Originally Posted by green80

Private design- not commercially available. I actually bought it from someone who said he couldn't win with it. I got more selective with it and it has produced a +10% roi over about 600 races bet.

Ok, I didn't know who you were, by your screen name, I thought you might be one of my black box owners who finally decided to do the record keeping analysis I suggest.

Speed Figure · 08-23-2015, 06:13 PM

Quote:

Originally Posted by raybo

That would be very little better than playing a single race. 22 to 24 races doesn't even account for a drop in the bucket regarding the long term. Variance alone, one large payout, one big loss, etc., could determine the winner of that small sample.

Well this whole thing is nothing more than "I'll huff and puff and blow your house down".

raybo · 08-23-2015, 06:19 PM

Quote:

Originally Posted by Speed Figure

Well this whole thing is nothing more than "I'll huff and puff and blow your house down".

Exactly!

But, keep in mind, their methods of attaining profit are fundamentally different. It would take months (and maybe not even then) to find out which one makes more profit. But, it would still be entertaining!

whodoyoulike · 08-23-2015, 07:08 PM

Quote:

Originally Posted by raybo

We had a black box challenge here few years ago, if I remember right it was over 6 different tracks over 6 days, betting every race on the cards, one win bet per race, betting the same amount on each play, best ROI wins. Handi's program came in first at 1.26 I think, and mine came in 2nd at 1.24 (after going 0 for 9 at Aqueduct). While the contest was entertaining, it hardly proved anything about any of the programs, or the other category of non-black boxes that were entered. Methods of playing, strengths and weaknesses, vary with the software/individual, so the chance of any single contest, over specific tracks, in a relatively short period of time, being equally compatible with all entrants are slim indeed.

This must have been before my time because I don't recall. We're not discussing a "black box" challenge but rather a .....

Mano a Machino or it could be a Womano a Mano a Machino or

some variation of a challenge. I still find the concept interesting and feel it would be both informative and entertaining. After all, I've now got all this popcorn ready to be microwaved. I don't think it should be handicapping every race because ..... Who does that?

Dave Schwartz · 08-23-2015, 08:30 PM

Quote:

I don't think it should be handicapping every race because ..... Who does that?

A computer handicapper with a successful approach.

raybo · 08-23-2015, 08:34 PM

Quote:

Originally Posted by whodoyoulike

This must have been before my time because I don't recall. We're not discussing a "black box" challenge but rather a .....

Mano a Machino or it could be a Womano a Mano a Machino or

some variation of a challenge. I still find the concept interesting and feel it would be both informative and entertaining. After all, I've now got all this popcorn ready to be microwaved. I don't think it should be handicapping every race because ..... Who does that?

My point was that, that contest, which also included non-black box participants as well, did not prove a thing except who had the best ROI over that period of time, at those particular tracks, and those particular races. There was one non-black box entrant, forget who it was right now, who beat both Handi and me, so it wasn't just machine against machine, there was human interaction by some of the participants also. Regardless, it proved nothing at all except that there indeed existed black boxes that produced well during that time period.

Speed Figure · 08-23-2015, 08:40 PM

Quote:

Originally Posted by Dave Schwartz

A computer handicapper with a successful approach.

(((((BOOM)))))

Leap · 08-24-2015, 12:55 AM

A computer longterm

Pick 'em Charlie · 08-24-2015, 06:29 AM

Selections are secondary. Most crunch the numbers similar. It's more about wagering.

pandy · 08-24-2015, 07:08 AM

Quote:

Originally Posted by Speed Figure

Why not play Saratoga and Del Mar next week? That would be about 22 to 24 races! that's more then enough to see who comes out on top!

I've been posting computer picks from my Diamond system on my website all meet for both tracks, picking every race except for 2yo races with firsters, which the system can't analyze. At both meets the top ranked horse is showing a profit. Now certainly humans have been able to show profits on every pick during a meet, but a good computer system is much more likely to produce a positive ROI than a human when a lot of races are involved, especially at difficult meets like Saratoga. A good human handicapper who shows a profit does it with spot plays. I've never known anyone who can bet every race and show a profit with any consistency.

garyscpa · 08-24-2015, 08:32 AM

Where do we get this Diamond System?

pandy · 08-24-2015, 09:10 AM

Quote:

Originally Posted by garyscpa

Where do we get this Diamond System?

On my website. Google diamond system handicapping.

classhandicapper · 08-24-2015, 09:42 AM

Theoretically you can program a computer to do almost any kind of analysis, but I could never program a computer to do the kind of subjective analysis I do in my day to day handicapping. It's not even conceivable. I could maybe get 90% there, but that last 10% is too human. Maybe some advanced AI gurus working with expert handicappers could do it, but it would probably cost a fortune to develop.

On the flip side, some of the studies and reports I have produced with my computer that can give me updates with a few keystrokes would take me years to do manually or to learn by trial and error.

I don't think it's an either or.

The data I can get from my computer makes me a better handicapper but my human qualities makes me better than my computer.

What the computer can do better than me is process large quantities of data.

If the contest was handicap 5 races a day, I'd beat my computer.

If the contest was handicap 200 races a day, I'd pass out and the computer would be done in a couple of minutes .

pandy · 08-24-2015, 10:17 AM

One of the reasons why some black box systems fail are the same reasons why many handicappers fail. Too much weight is put on factors that don't really matter. I've found, for instance, that if I include jockey and trainer statistics into my handicapping I will pick more winners. However, if I eliminate jockey and trainer completely, I will show a higher ROI. This is true of many factors. Average earnings per start may help you pick a few more winners, but could bring down your bottom line on ROI.

Most dirt races are won by horses that have enough ability to win the race. They figure to be relatively close to the pace (or on the lead), and have proven that on one of their best races, and within the margin of error, they can run fast enough to compete against the other horses in the race. Any horse that fits that profile is a contender. At that point, the handicapper has to decide which contender on ability is offering the best value.

In my opinion, based on research and testing and 40 years of studying handicapping, all of the other factors, pedigree, purse earnings, in the money finishes, jockey, trainer, workouts, trips, etc., can help to pick more winners but are of questionable value in terms of generating a positive ROI. I know many will disagree with me, but that's my opinion.