Neural Networks

Quote from Indrionas:

You emphasize "explainability" of a market inefficiency. Then I have this question: can this "explainability" be based on something technical like "this market behavior/pattern consistently repeated itself for past 5 (or so) years, thus it is possibly non-random behavior which can be exploited as a market inefficiency for some time"?

I'm asking this because currently I put a lot of effort into model building. I'm still in pre-development phase and expect to finish work in a few months. I'm trying to build models based purely on technical analysis, i.e. based on price data only.
More...

Thank you for "getting my point".

Note I do not use TA terms like "pattern"...
But market structure terms like "inefficient"...
Because TA is a false world much like astrology...
That is heavily promoted by the Securities Industry...
In order to entice semi-intelligent people to trade and lose.

EVERY successful business has a COMPETITIVE ADVANTAGE.
And EVERY successful business understands EXACTLY WHY it has a competitive advantage.

The idea that inexperienced retail traders...
Running simplistic TA methodology...
Would have a COMPTETIVE ADVANTAGE over, say, Bright Brothers top 50 traders...
Does not pass the laugh test.

You can trade based on quantitative analysis of price data...
I have been doing it for 15 years...
But the starting point MUST be a not-too-well known "market inefficiency"...
That you are directly exploiting...
because you have a COMPETITIVE ADVANTAGE.

PM me and I will point you in a viable direction.
 
Quote from HoundDogOne:

I have a degree in Computer Science (1982)...
And 15 years as a full-time trader...
Manage a 7 figure hedge fund...
And do about 3,500,000 shares/month.

So I have DUAL professional expertise...
And might have a different perspective then someone with more narrow experience.

There is no hard information from the AI crowd in this thread...
It's just a bunch of claims...
And the website mentioned (markrttrak)... is a joke.

Generally...
The field of AI involves modeling the most advanced intelligent entity there is...
The HUMAN BRAIN.

The only way to accomplish this in trading...
Is to model the thinking process of a human Expert Trader or Market Maker.

Something like a CBR approach to build an expert system...
Makes sense...
Especially since well-designed CBR systems have produced near pro level Poker Bots.

But, IMO, using sexy neural nets or immunologic "learning systems"...
WITHOUT ANY ACTUAL ACCESS TO A HUMAN EXPERT...
Will produce very marginal results.

This is akin to building Medical Diagnosis Expert Systems...
WITHOUT ANY INPUT FROM A MEDICAL DOCTOR.

You are better off in the long run...
Spending a few years becoming an Expert Trader...
Than just dogmatically building sub-optimal AI systems...
Which will be competing against Human Experts running ** far more optimal ** systems.
More...

Impressive credentials HoundDog. What's your funds performance % in the past 4 months, and the past 12 months "shooting from the human expert hip" ?
 
Quote from markettrak:

I wonder about a person who toots his own horn in this manner.

Trading that many shares per month equates to more than 150,000 shares per trading day. Sounds like a lot to me.

My trading record is available for anyone to see. Go to my homepage and click Example.

Rich
More...

You are still spamming this forum. If you want to be a sponsor, then pay the fee. Otherwise, please stop abusing this forum seeking free publicity.
 
markettrak,

if you know statistics as you all claimed, please post all of your OLD prediction, NOT selective potion/example of yours.

Has anyone know a site/method consistently post their AI results(for years)?
 
Quote from rcanfiel:

You are still spamming this forum. If you want to be a sponsor, then pay the fee. Otherwise, please stop abusing this forum seeking free publicity.
More...
This guy's site had zero traffic...
Which has now soared to ** almost zero **...
Since he started spamming ET.

http://www.alexa.com/data/details/traffic_details/markettrak.com?q=

No one cares about selective results you post on your lonely web site.

Unless you have Audited Financial Statements...
Or you are registered and accountable to a legit Regulatory Body...
Do not expect to be taken seriously by adults.
 
Guys,

I think NNs can be used to assist in trade signal development, but not in the actual development of trading systems.

For example, lets say you have two proven systems that make money. You want to integrate the two systems into one system with one set of buy and sell signals to further improve your win ratio.

NNs could help you here, but the actual trading rules need to already be validated first. NNs could be used to identify subtle interactions between the two systems and issue appropriate trades rather than using simple rules such as if both are long go long, if both are short go short.

Runningbear
 
Quote from Batman28:

genetic algorithms outperform neural nets CLEAN OUT..
More...

???

Could you explain this?

Genetic algorithms and neural networks perform different tasks. Genetic algorithms are optimizers, whereas (most) artificial neural networks are classifiers or regressions.
 
Quote from rcanfiel:

I had many discussions with many people some years back. THe general consensus was it was fool's gold. biocomp was one example
More...

Wait: BioComp is a software tool, not worked-out solution. That's like saying "Using saws to make birdhouses doesn't work. Craftsman is an example."

Having a tool and using it to solve a problem are two completely different things.

None of this is to say that advocates of neural networks in trading can expect the rest of us to accept their claims without evidence either, but the literature records many successful applications of neural networks to other analytical problems.
 
Quote from johnpinochet:

My regular work involves machine learning and data mining.

At home I use a combination of NN's, GA's and Tree classifiers.
More...


If I may ask: What do you do for a living?
 
Quote from rcanfiel:

I minored in AI (machine vision, robotics, plannnig/understnading, LISP, expert systems, etc.) as part of an MS in computer science. Your credentials??
More...

None of this is relevant. For a claim that "neural networks can never be used productively for trading" to be proven beyond doubt would require an impossibly exhaustive experimentation. The most one could seriously claim is something like "[assertion]: no examples of neural networks aiding trading exist and [opinion:] none is likely to surface". Whether that assertion is true or not, I cannot say, but superlative claims that neural networks can "never" work are silly.
 
You must log in or register to reply here.
Back
Top