it is impossible to trade across multiple global markets just using the neural net in your head.
best,
surfer
Neural Nets and Genetic Algorithms?
- Thread starter AsaFce
- Start date
it is impossible to trade across multiple global markets just using the neural net in your head.
best,
surfer
One can never affirm that it is impossible for a machine because it is impossible for human for the dumb machine can compensate with its combinatorial power that's why Big Blue has finally beaten Kasparov at Chess although it is sure more dumb than Kasparov, if Kasparov had the same power it would perhaps continue to beat the machines for more hundred years 
. Now as I said you can have the best Super Cray and the best architecture of Neural Net if you don't feed it with the appropriate model it will give you the same silliness as someone who would combine Excel formulas by random .
. Now as I said you can have the best Super Cray and the best architecture of Neural Net if you don't feed it with the appropriate model it will give you the same silliness as someone who would combine Excel formulas by random .Quote from marketsurfer:
it is impossible to trade across multiple global markets just using the neural net in your head.
best,
surfer
they are not trading size.
best,
surfer
Rather than using NN and GA to build a black box prediction model, the real meanful application should be in data mining area to classcify patterns and internal correlation between equity dynamic behaviors.
Any above average traders (both pro and amateur) share a similar experience, that by looking the chart, some TA, FA data, especially those stocks you've been closely watching for a while, you know by heart how the stock would behave in react to pressure of large market and segment, so any abnormal sign would be caught quickly by your instinct. The only question left out is that how confidence you feel about the predicted movement, contral your emontion, and the meney management, which truly distincghish a successful trader from others.
One way to make the above instinct/expierence based trading judgement into a process is to train your brain with extensive trading excise and traditional TA/FA analysis, though more of this kind analysis may ruin your instinct more or less over time, which explains why education in trading doesn't play a significant role.
Alternatively, by bring data mining tool (possibly NN and GA based), you may be able to identify your instinct and automate the trading analysis in volume. Even better, it may be take your instinct based analysis to next level as the sheer size of data the computer can handle is more than human brain does.
Remember, all your brain can handle is data mining. If there is any "prediction", that is because your brain has done a good job by analyzing historic data (data here means everything).
Any above average traders (both pro and amateur) share a similar experience, that by looking the chart, some TA, FA data, especially those stocks you've been closely watching for a while, you know by heart how the stock would behave in react to pressure of large market and segment, so any abnormal sign would be caught quickly by your instinct. The only question left out is that how confidence you feel about the predicted movement, contral your emontion, and the meney management, which truly distincghish a successful trader from others.
One way to make the above instinct/expierence based trading judgement into a process is to train your brain with extensive trading excise and traditional TA/FA analysis, though more of this kind analysis may ruin your instinct more or less over time, which explains why education in trading doesn't play a significant role.
Alternatively, by bring data mining tool (possibly NN and GA based), you may be able to identify your instinct and automate the trading analysis in volume. Even better, it may be take your instinct based analysis to next level as the sheer size of data the computer can handle is more than human brain does.
Remember, all your brain can handle is data mining. If there is any "prediction", that is because your brain has done a good job by analyzing historic data (data here means everything).
I think people like the idea of using neural nets and GA's becuase it all sounds cool and cutting edge. The reality is that no one has been able to build anything capable of making trading decisions better than a trader.
Sure some groups using them to optimise systems within a technical framework, but without a tight set of rules, they tend to start trading just like a new trader would. Poorly.
Better to stick with straight mechanical rules for simplicity.
Runningbear
Sure some groups using them to optimise systems within a technical framework, but without a tight set of rules, they tend to start trading just like a new trader would. Poorly.
Better to stick with straight mechanical rules for simplicity.
Runningbear
I'm going to demystify this stuff because there is some huge myth due to the expression "Artificial Intelligence" whereas at the current stage it is rather DUMB ! In fact among scientists AI had lost a lot of credibility until very recent years (until the 1985s). Only the mass public continue to dream on the moon about AI surely because of science fiction films. In the far future perhaps - advances will probably more come from Neurobiologists than from pure AI (although BioInformatics has begin to exist) - but today it is very pretentious to talk about "Intelligence". For example the first time AI lost its credibility was that some demonstrated that it couldn't even solve a simplist XOR problem (exclusive OR) ! Then hidden layers appeared to do that. But it didn't still do any miracle and scientists were still looking at ANN (Artificial Neural Net) as toys. It is since the 1985s that guys unified the ANN field with non-linear statistic field that NN has regained some seriousness. I have met a laboratory in France who is directed by the Pope in France of ANN and who has an international reputation - I remember that but I don't remember his name and his very point of view was that ANN was no more than statistics and that AI is a bullshit as for application to "Artificial Intelligence" per se. I didn't meet him personally but his direct associate. In fact it is because I knew his son who introduces me to him whereas I had already rejected that choice. They were collaborating with banks for using ANN and showed potential applications but I knew already about the bank models and for me the models were primitive at that stage and also it was not worth because I have a direct mean to find a white box model instead of using a black box model and they admitted it that it is sure far better to have a white box model. I would have been using ANN for parameters of the white box model if there were too many but it wasn't the case. In the future I will use AI for validating and find more rules of multiple trading systems based on this model (so that I can create a specific system for each type of trader's style) but as I said AI is not restricted to NN and for classifying patterns there are other techniques of AI than NN that use also GA as objective algorithm and which are less "black box" than NN - that's the reason I bought Edelman's book on Brain Conciousness model because Conciousness has much to do with Learning Process.
I intend to make a tutorial on the use of NN with application to my model and to indicators so that people will learn more easily how a machine "thinks" and if a stupid machine can do it there is no reason they can't do it also at least for simplest cases. As I used to make comparison with Chess training with a computer can improve your learning path.
and his very point of view was that ANN was no more than statistics and that AI is a bullshit as for application to "Artificial Intelligence" per se. I didn't meet him personally but his direct associate. In fact it is because I knew his son who introduces me to him whereas I had already rejected that choice. They were collaborating with banks for using ANN and showed potential applications but I knew already about the bank models and for me the models were primitive at that stage and also it was not worth because I have a direct mean to find a white box model instead of using a black box model and they admitted it that it is sure far better to have a white box model. I would have been using ANN for parameters of the white box model if there were too many but it wasn't the case. In the future I will use AI for validating and find more rules of multiple trading systems based on this model (so that I can create a specific system for each type of trader's style) but as I said AI is not restricted to NN and for classifying patterns there are other techniques of AI than NN that use also GA as objective algorithm and which are less "black box" than NN - that's the reason I bought Edelman's book on Brain Conciousness model because Conciousness has much to do with Learning Process. I intend to make a tutorial on the use of NN with application to my model and to indicators so that people will learn more easily how a machine "thinks" and if a stupid machine can do it there is no reason they can't do it also at least for simplest cases. As I used to make comparison with Chess training with a computer can improve your learning path.
Quote from harrytrader:
Your question is similar to "is anyone here successful using excel or a trading software ?". The answer would be yes and no
ANN is just a tool like Excel except that it is not a piece of software but a tool for modeling. But for modelling something you must know the parameters and the law of the world you want to model. And so there can be many hypes behind ANN sometimes. If you want to have an idea of ANN applied to stock market I just quoted a book in another thread "Neural Networks for Economic and Financial Modelling" Thomson Publishing (the edition I got is 1996 perhaps they have updated I didn't check) which deals also with genetic algorithm since these algorithms are used in the ANN for optimising the objective function. This book deals essentially with Feed-Forward with Back propagation model as they said that it is in their opinion the best suited general architecture for financial modelling.
Personally ANN has not been useful for creating my model. It is only after my model has been created that I could use ANNs profitably but I prefer another approach which is similar to what is called classifier (but at the moment there is a practical problem of databases). Although it has the capability of being statistically an Universal Approximator for non-linear model - and my model is non-linear - notwithstanding that the structure of the network is also correct - this is essentially a black box modelling approach and I prefer white box modelling approach because you model the causes and know what you are doing by design whereas with a black box approach there are much uncertainty about the robustness of the model.
It's an application that runs on your pc.
I've been playing with it for a couple of years now-- I've generated a few models that were profitable, but not tradeable (for me).
The people who have done really well with it knew how to trade already, and neuroshell took them to the next level.
It's also extremely useful for systems testing (even without the neural net bits)-- I use it all the time to backtest some of the systems discussed on et and other places.
Anyway, if you've got the $Xk to spend, it will be a worthwhile journal of learning, but you'll be disappointed if you think its going to be the magic bullet.
regards,
laziz