My quest for the holy grail

Quote from braincell:

Ok so look at it this way. you had idea1, idea2, idea3, etc. Each idea had technique1, technique2, technique3 etc in it. All of those techinques and approaches were glued together in method1, method2, method3.

Now. Make the computer jumble up methods, techniques and ideas, and make strategies out of them. Make it do that and backtest/fwdtest all of them (millions) until one comes up. Take out the creation and assumption part out of your strategy creation attempts and let the computer bash it out. Maybe it will find something. Humans need to be replaced completely, that's the problem.

I was in the same position you are in now. Do not dispair, keep trying. This is a learning process. You pay for education.
More...

not despairing, I don't trade for living and it's much more of a hobby than an obsession. I'm not trying to take humans out of the equation, it just seems that humans are so full of shit, you couldn't really conclusively tell if a trader is good or bad or having a good year or a bad year, so looking for order in chaos
 
Quote from Mr_You:

I personally don't believe in patterns. Only what the data/market is telling me right "now". Smoothing the data helps reduce the noise, makes it easier to work with.
More...

if you don't believe in repeating patterns then how do you base your decision on what to do with your "smoothed data"? I don't nessecarily believe in technical patterns but I do believe that most people will react in a similar way when they are afraid of losing money or afraid of missing the boat. but haven't found anything conclusive yet, so it could all be wrong, it could be totally random and the only people that are consistently profitable are the assholes on forums who already know everything and have since given up any creative thought, when they win, they win, when they lose, it's because the market is acting funny or something, lol...
 
"...scour for patterns..."

A fruitless never ending quest for something which does not exist. The true quest is to find what the now bar must do as it evolves from the then bar.
 
Quote from Mr_You:

I personally don't believe in patterns. Only what the data/market is telling me right "now". Smoothing the data helps reduce the noise, makes it easier to work with.
More...



.......................................................................

damn.....

i have been trading this stuff for 35 years that does not work......

why was i not told earlier.......?

the market is fractal.
get your head wrapped around it

the patterns continue in all time frames and the game is all math.

i especially hate when a DT setup hands me 140 ticks for a few hours work.

cheers,

s

http://www.elitetrader.com/vb/showthread.php?threadid=223293
 

Attachments

Quote from elliotwave:

care to elaborate, I would like to get your point
More...

Make an application that creates strategies from blank and tries them out. Have as little assumptions as possible. Backtest huge numbers of them, and make sure your trades per parameter ratio is low. Basically that means the entire process of the "quest" is left to the computer. You should no longer be doing the quest, but make AI that is doing the quest for you. It's a cutting edge concept and I'm not sure I should be talking much more about it. It works for me anyway. It works very well indeed. Does that help?
 
Quote from braincell:

... Make the computer jumble up methods, techniques and ideas, and make strategies out of them. Make it do that and backtest/fwdtest all of them (millions) until one comes up. Take out the creation and assumption part out of your strategy creation attempts and let the computer bash it out. Maybe it will find something...
More...

the above = "curve fitting"...

The important result is not that the computer manages to find some "good results"; it's that it has to discard "millions" of alternatives to get there. "Survivorship bias"...
 
Quote from abattia:

the above = "curve fitting"...

The important result is not that the computer manages to find some "good results"; it's that it has to discard "millions" of alternatives to get there. "Survivorship bias"...
More...

Hehe. That kind of thinking is so 2010. It's nice to read stuff like this. Makes me surer I'll keep my edge for a long time. :)
 
Quote from abattia:

the above = "curve fitting"...

The important result is not that the computer manages to find some "good results"; it's that it has to discard "millions" of alternatives to get there. "Survivorship bias"...
More...
If you believe the story, I guess you could say Edison curve fit his light bulb design... I know what you mean that you can make data do anything you want, but it probably won't work with another set of data, anyone that only tests with one set of data is short sighted
 
You must log in or register to reply here.
Back
Top