The term AI has been overhyped. It ressembles nothing of really artificial intelligence at the moment. In fact NN has lost much of its credibility because of this hype (for the scientists, for the public hype still works) so that now some specialists of NN - I met a few years ago a famous one he was an international specialist and his lab was working on models for petroleum and financial domains - even refuses the use of AI term and prefers to talk about non linear statistical approximator. Apart from that I have nothing against. As for my own use I don't see much utility: I prefer my own neurons not that I trust them but I can conserve full understanding of the parameters of the system and avoid overoptimisation and fuzzy logic. Now this is my case I don't generalise for others.
harrytrader, I totally agree with you that AI was hyped as a term, and I remember how it was almost taboo in the late 1980's. Still there were persons continuing doing work in the field, and that's why we have more diversified options today, like Kohonen maps etc. I also use my own neurons to do my trading - and will continue to let them have the last say. Still, I find that I do trading based on not an infinite state of mind, nor infinite variations of signals. Sometimes, however I lose my concentration, and do not process those same signals as well as I do other days. Therefore I am really looking for some confirmation and continuation in identifying some of the indications I do on a daily basis. Thus a more sophisticated means of compiling indicators/signals would help me, if I can have some of them modeled precisely. Then there are other contra-indicators and signals that I would use to make trading decisions. When I use some pattern for basis on my trading decisions, I sometimes have a very short timespan to close the trade, and these times I would rather have an automated trigger help me, because I know I sometimes am not focused enough to consistently adhere to my current system/rules. I imagine some mixed system for this, using rule based logic, and then more probabilistic learning methods to try and model some of the indicators I (think I) "know" already. Rule based logic can be programmed in procedural languages, but much more elegantly in declarative languages. It's just like Lisb and other functional programming languages is way more efficient at e.g. list processing like time series etc. I remember a language called Clean, developed by Univeristy of Nimegen (kind of Miranda-like, Haskell) with lazy typing and quite efficient performance. Those kinds of ease-of-expression tools/languages does help me model things more efficiently, and therefore I look at things like Prolog etc. for easier prototyping - just like I used Beanshell to very quickly deisgn my current trading/analysis platform. Now, it seems that HMMs are quite good at recognizing sequences of states, but won't necessarily generalize well. This might be a good thing if the markets are constantly changing, in my opinion. Likewise, I have seen promising features of Bayesian Networks - as well as my new found interest in these kinds of networks - which I'd like to explore some. They seem to model well on high frequency data, unlike NNs which would prefer very normalized data, with extremes filtered out etc. NNs are better for generalizing, although I am reading how BNs and HMMs also can be made into generalizing by using median terms. If one used one's own trading decisions as one of the inputs, how would this possibly map out in some of the models .. could the model quite accurately map your decisions over some timeframe. With "mood" data also entered, would it be easier (nervous day, anticipation, bearish, bullish etc. based on news and other data) ? Could a model advise you when not (!) to do a trade based on previous failures ? It's interesting, and probably some highly individual results could be obtained. With regards to Kohonen maps (SOM), I'll look into that too eventually with the JOONE library - http://www.jooneworld.com , and see if a model for me is feasible with this library. Besides, all the software and libraries I am using are free.
Of course there are still research in the field. In fact there is even a renewal because of interdisciplinarity: NN is no more investigated only by pure programmers but neurobiologists now entered the field. As for automating trading decision I also of course agree with you since that's what I'm doing (see http://www.elitetrader.com/vb/showthread.php?s=&threadid=30651). Why I do that ? Because I know perfectly what is the burden of a daytrader: being able to process quickly information. I even said here : http://www.elitetrader.com/vb/showthread.php?s=&threadid=14637&highlight=refraction 'There exists a "psychologic refraction period" that is necessary to treat any elementary token of information. It is at least a few hundreds of millisecond. And our brain is like Windows : it is a false multitask processor but only monoprocessor. So all informations are treated sequentially and take an important time to process. That's why in daytrading, above all at sub level scales, one can be stressed if not prepared enough because too much information to treat at high speed.' P.S.: you can also see that I am not uninterested by AI per se since I said "When I was 12 I wanted to become an astrophysician. At 16 I wanted to become a neurobiologist. " I just say that it has been a hype and you also agree. P.S.2: for my automation process above, we've in fact developped something that belongs also to AI: an expert system engine. Nevertheless I just don't publicise on that as well as I rarely use the fact that my model belongs to non-linear dynamics or Chaos Theory field because I don't like to use fashionable terms for the public. I don't want to cope with people that are just impressionable by these terms. For me they are tools no more or less than Excel is a tool and Excel is not a substitute for my Knowledge and Experience (K&E): I use these tools as a Container of K&E not as a Creator of K&E. Finally we agree in everything that's not fun : usually I search for an opponent to what I said ... just to rehearse myself .
Quote from LongShot: lol LongShot keeps everybody communicating. Thanks for the link on HMM. Here is some of what you seek: http://www.elitetrader.com/vb/showthread.php?s=&threadid=9437&perpage=6&pagenumber=10 It's a bit spread out, but; it is all there using the daily time frame.
This thread is another example of why ET is so great. A quant post started by Gringinho, which leads to the great debate, which ends up producing links and information that I have never heard of. Everytime I think I'm getting somewhere, I find out I have much farther to go. I just started doing testing for intraday high and low and what time they occur. I am setting up testing code which does a distribution test so I can find the probability of the time of high and low. And now to see that incredible article which addresses this concept in great detail, it opens up many new testing areas and perspectives. I love trading and the discovering and all you guys and this site just add value to the whole process. thanks to all..... I'm still debating whether to look through all of Grob109's posts or not? I don't like getting something for nothing, I like to research it myself, but it is tempting .... Also, I just started using this new browser "Avant browser". It's free. I like the way it keeps tracks of the links you have gone to at the bottom of the browser. You can just click on a link and you are taken right back. It's a time saver for doing research. http://www.avantbrowser.com/ XT
...I presume that once you determine those most probable high/low times you may want to short the high and long the low. I advise you to run a companion test to determine the expectation for shorting ANY price in a rising market or longing ANY price in a falling market. You might be surprised at the results.
I'm not interested in Price patterns, I'm interested in psychology of them: - trading technics rather then crystal balls! Cheers, its always fun to read your comments.