Neural Networks

Quote from Indrionas:

There is one thing I don't understand when people say "find an edge". I can think of two possible "edge" types. One is fundamental and other is technical. I imagine fundamental edge as something based on fundamental, relatively subjective data, like investigating company's results before buying its stock, something like that. Technical edge is based on technical and/or statistical analysis of price data, like "if condition X is present, then there is probability P that event Y will happen". So when one says "find an edge", what does he mean? Statistical significance by objective analysis or some obscure subjective analysis (i.e. "gut feeling", intuition, discretion, etc.)?
More...

Let me explain.

If the markets were perfectly "efficient"...
It would be impossible for anyone to profit... (without cheating).

But the markets have countless "inefficiencies"...
And 1000s of trading operations are consistently profitable.

ALL profitable operations exploit a specific "market inefficiency".
Repeat ALL.

If you cannot explain the "market inefficiency" that you are exploiting...
Why is exists...
And why you are able to exploit it...
In one minute or less... or 100 words or less...
Then what you are doing is worthless in the long run.

This is basic logic.
You cannot circumvent basic logic and be a successful trader.

Your distinction of fundamental vs technical vs objective...
Is artificial and not useful.
Unless you stop buying into the Securities Industry con game...
And their clever doublespeak...
You are doomed.

Ignore 100% of the crapola the Securities Industry and Media feed the mooks...
Because they have ONLY ONE reason to exist...
To take your f*cking money.

In summary...
Whatever method allows you to consistently find and exploit "explainable market inefficiencies" is legit.
 
Quote from HoundDogOne:

Let me explain.

If the markets were perfectly "efficient"...
It would be impossible for anyone to profit... (without cheating).

But the markets have countless "inefficiencies"...
And 1000s of trading operations are consistently profitable.

ALL profitable operations exploit a specific "market inefficiency".
Repeat ALL.

If you cannot explain the "market inefficiency" that you are exploiting...
Why is exists...
And why you are able to exploit it...
In one minute or less... or 100 words or less...
Then what you are doing is worthless in the long run.

This is basic logic.
You cannot circumvent basic logic and be a successful trader.

Your distinction of fundamental vs technical vs objective...
Is artificial and not useful.
Unless you stop buying into the Securities Industry con game...
And their clever doublespeak...
You are doomed.

Ignore 100% of the crapola the Securities Industry and Media feed the mooks...
Because they have ONLY ONE reason to exist...
To take your f*cking money.

In summary...
Whatever method allows you to consistently find and exploit "explainable market inefficiencies" is legit.
More...


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.
 
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...

I think what HoundDogOne is getting at is something like the difference between the following propositions:

A. When the price of certain stocks deviate "sufficiently" from VWAP there is a tendency to mean revert.

B. When the 20SMA crosses above the 50SMA, it is a buy signal.

There are plausible reasons for the first which lies in the volume of VWAP based trading which is not in question. Incidently there is some truth in this proposition.

There is little explainablity about the second other than the very weak assertion that lots of traders watch these MAs - an assertion that is impossible to verify.
 
Quote from markettrak:

I use genetic algorithms to generate neural networks for trading. The combination works very well.

Rich
More...

May I ask you to elaborate a bit more?
 
Quote from 9999:

May I ask you to elaborate a bit more?
More...

Typically you set your NN's 'predictive' or 'sorting' capability to be the fitness function, and the weights of the edges to be the genes. Things get a bit hairy, however, if you have chosen the wrong sigma functions, or the wrong number of hidden layers. You could indeed figure out a way to throw those in the genetic pattern if you wanted ... but your search might take a while :D
 
Quote from Craig66:

I have to ask the obvious question, if it works so well, why bother selling it?
More...


I believe he's selling the signals only, not the system.
Markettrak, thank you for the explanation. Do you use a proprietary software or a commercial one for generating GA and neural nets?
Thanks again.
 
AI is only a tool. If you know how to used it, it might help you making money; if you don't know how to use it you will never make any money with it.
Never think that AI on itself will make you money.

I know someone who makes good money with a system build in Excel. Does this mean that buying Excel will make you money? No; the same logic applies to buying an AI system.

If you have a really good system that makes lots of money, you will never sell it. Unless you get a huge amount of money. All the others reasons are b*llshit.

I bought an AI development kit years ago. I asked someone with an university degree in computer science and mathematics to help me set up the input for a trading system. We never succeeded. The fact that the guy had so much problems in setting up something decent made me realise that what i bought was a tool, nothing more than a tool.
Perhaps the guy was an idiot? I don't think so because of his degrees. The guy was "headhuntered" in Europe and is now in the top of Reuters USA.
 
Quote from markettrak:

That’s a very good question. I think you could ask it to any of the vendors selling market forecasts. For me, as a physicist, I want a forum to show that you can forecast the market with reasonable accuracy. I could publish a paper on the model but then I would be giving it away. This model is far from perfect but I do make money trading the signals. The subscription fee I charge is just enough to buy new computers and pay other expenses.

Rich
More...

I got his signals for a long time during several different intervals. They don't work. He just believes they do.

And you are violating the terms of this site by promoting your website without permission. So please get lost.
 
Quote from rcanfiel:

I got his signals for a long time during several different intervals. They don't work. He just believes they do.

And you are violating the terms of this site by promoting your website without permission. So please get lost.
More...

I agree with this.

His ludicrous claim "Average Accuracy: 80.3 percent (20+ years)"...
Is worthy of a 19 year old.

If you're gonna con people with fuzzy AI references...
At least pull up your socks.

Or maybe start a sports tout service using the AI angle...
Degenerate sports bettors are even more gullible than degenerate day traders.
 
You must log in or register to reply here.
Back
Top