Negative Class Drops and Handicapping Magic

[mhrussell]

I wanted to solicit ideas for some general rules to code up into my home grown program regarding negative class drops. I use the Handicapping Magic techniques, so a class drop is considered at least “1 reason” to go past the current pp line (2 reasons if the class drop is 2 levels or more). I am finding that all too often, some negative class dropping horses are getting through my FCW analysis without being noticed and I am using a much better race to base the horses expected performance than is justified. What I want to do is program a flag that will alert me to a potential negative class dropping horse so I can look deeper at the pp’s and decide for myself what pace line/form cycle window adjustments need to be made (if any) for that horse.

So, to this enlightened group, I ask the following three questions:

1)Can we specifically define what a “negative class” dropping horse is? Or is it similar to the definition of pornography: “I can’t define it but I know it when I see it”.

2)What set of rules/conditions can be developed to capture our definition in 1) so it can be included in a computer program? Again, my idea is to program an “alert flag” which would indicate a horse is a potential negative dropper that should not be considered as a contender.

3)I’ve read several recent articles noting that more negative class dropping horses are winning races than in the past, and that this is an unexpected and interesting development in racing. How should this fact impact our handicapping these kinds of horses?

I know the public usually likes horses dropping in class and tend to bet these seemingly “classy” horse. But sometimes the public too is suspecious and the horse does not get the action one would expect. This begs the question, when can we bet this horse? How should we consider betting action in the mix of all of this?

Any comments or suggestions are appreciated.
Thanks,

[Lefty]

Personally, I view a negative class drop as any winner dropping 1 or more levels. An in the money or horse within 3 lghts of winner and dropping 3 or more levels I also view as negative and I love it when this kind is favorite. They don't win their fair share.

[SAL]

Also, a horse that drops below a price that it was claimed for. I'm not sure if you can program this, however. And any horse coming off an extended layoff with a steep class drop is a definite negative to me.

[TonyK@HSH]

The nature of a negative class dropper has changed as purses rise due to slot machines. Most owners are in the game to have fun and make money. Now days, at certain smaller tracks (CT, DEL, MTR) an immediate class drop is not a negative but a shrewd business move. For example, at MTR open $6500 claimers run for about $21,000. This means more than $12000 to the winner. If I claim a horse for $8000 it makes perfectly good business sense to enter back for $6500. I risk a $1500-$2500 loss for a quick $12000 payday knowing that I'm often running the best horse on the drop.
I'd suggest that once you identify the potential 'negative drop' scenarions you review them as some may not be as bad as they appear on the surface.

Good luck

Tony

[so.cal.fan]

As a general rule:
let good sound business be your guide.
If YOU owned the horse in question..........and it was in reasonably good condition.......what would you run it for?
Be offensive.......not defensive......but be reasonable.
If a drop seems too severe.......it probably has hidden negatives.
But, again, keep in mind, how much does it cost an owner to have a horse in training? What have you put into the horse?
How much did you pay for it?
Ask all these questions and then put a fair market price on the horse.

Now, just how you might program this into software.......I don't have a clue.:confused:

[TonyK@HSH]

Good post So.Cal.fan,

Common sense is the best guide and get familiar with the owners/trainers on the circuits you play. Some are aggressive and try to strike for a quick payday. Others are leary about losing a 'pay' horse. This one may be tough to quantify in a software program. One thing is for sure, class drops like this will attract attention at the windows opening the doors for fair priced winners. Remember if the most well meant horses lose.

Tony

[SAL]

All these factors are reasons why it is always best to pick your own pacelines.

Over the weekend the program I use picked a paceline 3 races back. The last two races were inferior to the paceline that the program picked. What the software did not note was that the trainer of note for the paceline race was different than the trainer today, and the difference was substantial.

But anyone blindly playing computer picks would have saw the final rankings and stamped the horse a contender using the default picks. Needless to say, the horse ran out of the money.

That's why programs themselves will not beat the races, human insight is sometimes needed. I myself do not bet unless I am the one who picked the paceline for the race in question.

I actually like the concepts of Handicapping Magic and the ideas that Pizzola presented in the book. I got a lot out of that book.

SAL

[andicap]

Excellent post, Sal!!!
I've been saying for years that software programs that use models based on computer paceline selections will go nowhere. Believe me I tried one a few years ago and never made any money using it. Models would NEVER stand up going forward.
That's a great example of how the computer paceline will lead you astray.

But here's something interesting to think about: Over at HTR they are running a contest -- the automatic PL of the program against handicappers picking their own lines.
Although they only have a small sample done, guess who's winning?

The computer! Maybe this will change in the long run -- in fact, I suspect it would -- possibly longer after the contest is over. But I'm still surprised by even the initial results.

[Lefty]

I'm not at all surprised that, overall, the computer will pick pacelines better than a human. Reason: Computers are not swayed by emotion, indecision and personal prejudices.

[GameTheory]

A human expert is probably able to pick better pacelines than the computer, but humans are not consistent. You may pick one paceline over another because of what you had for breakfast.

One thing I find curious is that people are always making comments like, "A computer will never be able to do this, or a computer will never be able to do that." but the things mentioned are usually things that could be programmed into a computer if anyone cared to. For instance, in Sal's example the computer apparently did not check/care whether the trainer 3 lines back was different. It would be no big deal to add that to a program. Most of the things that people talk about are not things that have been tried and failed with a computer, but things that have never even been attempted, at least with commercially available software.

I'm building my own system, and I'm not aware of any program that attempts to account for all the stuff I've got in there. If I used *every* available program simultaneously they still wouldn't cover everything. All commercial software that I'm aware of is basically a "lite" version of what is possible, probably because you couldn't hope to sell such a behemoth. This is due more to the fact that the massive amount of data needed is not freely available than any programming difficulty. Lots of people actually say they don't want to take into account too many factors because they'll be over-handicapping or will just end up emulating the public. I agree with that when you the person are doing the handicapping, but my philosophy when making software is to use the computer for the things it is good at. If you write a program that does the same things you can do, only faster, you'll probably always be able to do best when adding your own judgement to things like paceline selection. There will always be shortcuts required when programming the computer to emulate your method. But if we use the computer to do things that *only* a computer can do -- data mining, modelling, AI-type prediction, etc. -- then you're really adding something new to your arsenal rather than just speeding up the whole process (which is also useful, naturally).

I seem to have gone rambling there... sorry.

[MarylandPaul@HSH]

I'd agree that it's difficult to build a rules-based paceline system that works for all races, but I think the chances for long term success go up substantially if different systems are used for different types of races.

I don't use the same paceline system for older claimers as I do for 3yo NW-1 Allowance races, for example.

MP

[keilan]

Very true indeed, absolutely key advice -- and you managed to say it in two sentences.

[GameTheory]

Originally posted by MarylandPaul@HSH
I'd agree that it's difficult to build a rules-based paceline system that works for all races, but I think the chances for long term success go up substantially if different systems are used for different types of races.

I don't use the same paceline system for older claimers as I do for 3yo NW-1 Allowance races, for example.



Exactly. But there is no reason whatsoever that you wouldn't program your computer to use different methods for different situations. I think many people believe a computer program automatically means a "one size fits all" approach, which it doesn't...

[SAL]

GT,

I absolutely agree with everything you said. I am well aware that computers can be programmed to do alot of things. In fact, HSH has a custom paceline selection option that let's you build your own paceline picker. You can build several for different situations.

But I don't WANT to have the computer do everything for me. One critic before the contest said that software users had no business competing. As if using a computer to assist your analysis of a racecard was criminal or something. I like having a program do what you just said, for numbers crunching and data analysis. In the end though, I use handicapping principles and judgement to make my selections and bets, not the computer.

I really enjoy handicapping, the challenge, and when I'm right, the financial rewards. I don't want to automate all of the process to where I just look at the top rankings box the top three choices in an exacta box.

The contest was a fun thing for me, I had never entered one before and I enjoyed participating. But I used a lot of personal judgement in my selections, and I got extremely lucky. But even without the monster day I had, I am proud of the fact that my top choices showed a decent plus ROI, which was my goal before the contest started.

I still believe that the program that can beat the races without human intervention has not been written, and may never be.

SAL

[GameTheory]

Originally posted by SAL@NETCAPPER
GT,

But I don't WANT to have the computer do everything for me. One critic before the contest said that software users had no business competing. As if using a computer to assist your analysis of a racecard was criminal or something. I like having a program do what you just said, for numbers crunching and data analysis. In the end though, I use handicapping principles and judgement to make my selections and bets, not the computer.


Sal, I've followed your picks in the Player Picks section of netcapper.com (I assume that's you) and obviously you did real well in the contest. (Let's face it, you single-handedly won the thing.) You are a force to be reckoned with -- whatever you're doing is RIGHT.



I still believe that the program that can beat the races without human intervention has not been written, and may never be.


Well, I'm writing it, so we'll see. I wasn't trying to tell anyone that they should or shouldn't do anything a particular way, just noting what is *possible*...

[sjk]

GameTheory

I definitely agree that it is possible to develop a successful program that works without human intervention. Good luck with yours.