Better NN EA - page 5

Craig
235
Craig  
derikb:
I have to agree with MiniMe. I had spend a lot of time on NN and NN only curve fit your data. You can't predict but you can follow. NN better to show where price will not go than to predict where it will go.

All TA is curve fitting, future predictions based on patterns in the past.

NNs are just another way of looking for these patterns.

w4rn1ng
192
w4rn1ng  

Why not, why not predict the future!

It's impossible to predict the future, lol..

But, a well trained NN is not created to predict the future, i think all you are looking at the NN with a wrong attitude, a NN should be like a intelligence, like humans have, look at it as a entity, that can think, can calculate, and can choose.

Now, you get a NN, and train it, let it (he?) study the forex market (studying the chart), and let it study and study, when a certain time the NN say: well i understood everything, try me and i let you see that i've understand how price work (you query he by letting it doing some backtest).

Then ask to him (or to yourself): if you understand and understond how price work (worked in the past), surely you must have built a strategy, an algorithm, some thinking method that let you out of bad trades in these 5 years (5 years of backtest for example), now.. apply this strategy/algorithm with the future price, if you really got the perfect strategy (that worked well in the past 5 years), there are good probabilities that it will continue to work well for the next year!

And, it may work, for the next year, or for the next 10 years, or only for the next 5 days, i've worked with NN for some years, and i know that it can work, but at a certain time it just stop to work..

But i noticed that if a NN is trained well, and with a big amount of data, the results it give out are more (much more) then 50% right, neural network can work great! The only problem is how to integrate it with MT4, because i found that it is impossible to create a good NN with just mq4 code, and also to well train a good NN it is needed a too much powerful computer (compared to what i use), but it work.

Just my opinion

oilfxpro
442
oilfxpro  
granpa:
probability is % of chances and predicting is result of probability.

for example: if you put gun to me:

probability 1: gun failure : a%

probability 2: bullet not inside : b%

probability 3: shaking hand resulting miss his shot : c%

probability 4: someone come and help : d%

probability 5: bla bla bla : e%

and so on

so, prediction whether i will die or not is depend on % probability.

-

Now if somebody was really holding a gun at your head ,it would knock some sense into it

GET REAL !YOU GOT TWO CHOICES .RED OR BLACK .You don't know which one

IT COULD BLOW YOUR HEAD OFF ,

oilfxpro

oilfxpro
442
oilfxpro  

Hi

Another point

A casino owner knows his probabilities of winning.

He knows he is going to lose to some punters

According to you lot, he is trying to predict the days he will lose?bUT YOUR NN WILL TRY TO PREDICT THE DAYS IT LOSES

Ha Ha Ha

OILFXPRO

[Deleted]  
zupcon
313
zupcon  
oilfxpro:
Hi

Another point

A casino owner knows his probabilities of winning.

He knows he is going to lose to some punters

According to you lot, he is trying to predict the days he will lose?bUT YOUR NN WILL TRY TO PREDICT THE DAYS IT LOSES

Ha Ha Ha

OILFXPRO

Exactly. The casino operator has an edge, and the longer that edge is played, the more successful he will be.

I really don't know where this fallacy that technical analysis tries to predict outcomes comes from. Ive read 1001 TA books, and a 1001 TA websites, and on not a single occassion have I ever read that TA predicts. The best it can do is skew probabilities in your favour

And yet, the forums are littered with people who in some way think that TA is supposed to predict, where do they get this strange idea from ?

oilfxpro
442
oilfxpro  
zupcon:
Exactly. The casino operator has an edge, and the longer that edge is played, the more successful he will be.

I really don't know where this fallacy that technical analysis tries to predict outcomes comes from. Ive read 1001 TA books, and a 1001 TA websites, and on not a single occassion have I ever read that TA predicts. The best it can do is skew probabilities in your favour

And yet, the forums are littered with people who in some way think that TA is supposed to predict, where do they get this strange idea from ?

There are certain set ups in technical anylysis ,which show good outcome if positions taken in a certain direction but that is just about all you can use from t/a ..................probabilistic set ups combined with other factors

OILFXPRO

Craig
235
Craig  
w4rn1ng:

But i noticed that if a NN is trained well, and with a big amount of data, the results it give out are more (much more) then 50% right, neural network can work great! The only problem is how to integrate it with MT4, because i found that it is impossible to create a good NN with just mq4 code, and also to well train a good NN it is needed a too much powerful computer (compared to what i use), but it work.

Just my opinion

You can actually get reasonable results with very simple linear combinators which can be optimized in MT4, but I ended up writing my NN trainer in C++ as debugging MQL code is like kicking yourself in the head over and over again.

The trainer writes the NN results to a database which is read by a DLL used by the MQL script, so it's all automated.

oilfxpro
442
oilfxpro  

Hi Guys

Anybody post all the trades of better and some anylysis?

Maybe we could crack the egg

OILFXPRO

project1972
563
project1972  
oilfxpro:
Hi Guys

Anybody post all the trades of better and some anylysis?

Maybe we could crack the egg

OILFXPRO

The NN logic is not linear, Better EA sometimes trade Counter-Trend and Sometimes Trend-Following.

You can't back-engineering a non linear logic, most of the time not even the author know what are rules learned by a NN logic, it's a black-box for everyone including his own creator.

You can check the past trades, but you don't have a clue what rules will trigger the next.