http://www.BillHowell.ca/Neural%20nets/Neural%20nets/Howell - Financial markets and Computational Intelligence, questions.txt www.BillHowell.ca 10Nov2020 initial A problem with the questions below is that I'm not so interested in what interests me, but in how you see (Computational Intelligence (CI), AI, machine learning), where it is and where it is going, and how that applies to (markets, trading, etc) in the context of traditional tools of analysis. A few other general market themes are also of interest to me. But rather than going through the questions, pick and chose any that you would like to respond to, and more importantly, bring in issues that you see as more important from YOUR perspective! I did say at I will pay up to 200$US for your time, so we need to cap discussions at that. Bag of questions : pick one or whatever interests you : What do the terms "AI" and "machine learning" mean to you? Does it include "Computational Intelligence" (CI - Evolutionary Computation (including swarm particles, immune systems, etc), Neural Networks (including Deep Learning, Convolutional and recurrent NNs), Fuzzy systems)? Are you mostly focussed on Natural Language Processing (and probably "streaming Big Data" for (news, social media, etc))? Classical techniques - Lee Giles recently presented during the International Neural Network Society (INNS) online workshop on Explainable AI. He repeated a theme from his IJCNN Budapest conference that his work is increasingly emphasizing non-CI tools from [compiler systems, deterministic finite-state automata (DFA), and I think maybe Natual Language Processing (NLP)]. As usual, there is a [power, speed, robustness] of these tools where they can be applied. Do you have a sl way of blending disparate approaches? Without providing proprietary info (maybe just what does "OBV stand for?), what kinds of mathematical tools do you see as key to investments, both trading and mid-to-long term? How do you balance conventional and "AI/machine learning? Fractals, Elliot waves, Socionomics - Fractal levels - Are these just guide-posts you, or actively used in targets? What about timing? I've never seen Mandelbrot's multi-fractal type of approach for that? Elliot waves - I have to say that I've never been (competent, comfortable) with this, but it's important to me because it anticipates trend changes at all time scales. It seems to me that in many areas, trends and patterns along a trend are difficult enough, but the real issue is the anticipation of major changes in (state, phase) of a system (eg climate, markets, etc). This is where it seems everyone fails,the Elliot-Wavers may have been more successful than most? Do you it much? Do you use chaotic attractor phase cross-overs for changes? Or what do you have most confidence in? Socionomics - Whether (right, wrong, true, false) is not so important to me. Robert Prechter's books are thought-provoking and it seems to me that they at least attempt to address the severe limitations of much of the mainstream academic thinking, which has value in itself. I've only recently read through this, and was surprised to learn that Elliot was thinking along those line early on (~1930s?). He did point out social media etc might provide even early measure of social mood. Were you influenced by socionomists? Apparently your business started with (condensed, meta-level) analysis of media streams. Neural networks - There have been many neural network papers on mood and emaotion analysis, but I've not paid attention to this area, being perhaps insensitive to the sensitivities of others (and perhaps myself - important to fight count-productive tendencies with (trading, investing)? Trading patterns - There are a huge range of market patterns at all time scales. How much pattern analysis do you do or feel comfortable with? Not just the head-and-shoulders type of stuff, but going well beyond that. Here is a strange point, although my main hobby has been neural networks science since 1988, I haven't actually focussed on them for patterns, even though they have of course been very uccessful for this. Risk analysis and Trading insurance - - Your course outline describes this, and it may be too much to get into without my doing much more homework on it (unlikely). I've only rarely used puts in the past as an insurance (I'm a very timid investor), but they aren't cheap even when they are cheap, and beyond an insurance role they may not be safe for lousy investors such as myself (why leverage bad peerformance?). Do you regularly apply them, or only in special situations? I won't even ask about shorts. I've only used them a couple of times, and was lucky to not run into the dark side of them. I assume that information of (futures, shorts, call, puts) is handy in your analysis? Financial system risk and changes to more direct government intervention - I'm not comfortable with more government "control", but that seems to be the trend, and the desire of the electorate. To me, "Social policy manipulation of the marketplace" may be inevitable to some degree, but dangerous (eg Modern Monetary Theory, massive (Treasury, Fed) interventions). Do you account for any (systemic changes in markets, risk) on this side? Are perhaps shifting to a higher long-term rate of financial asset inflation, as may have happened after the Federal Reserve was first set up (1913 or so?)? Part of this question relates to : http://www.billhowell.ca/economics,%20markets/SP500/multi-fractal/1872-2020%20SP500%20index,%20ratio%20of%20opening%20price%20to%20semi-log%20detrended%20price.html Macro-economic analysis - I suspect this isn't relevant to a short-term trading perspective, other than primary data releases etc (employment, GDP growth, etc) and how people may react to that? But do you still use it for a "reference ground truth" as an indication of risk? Evolutionary Computation (EC) - While my focus is on Neural Networks, I'm a huge fan of EC, and cer it to be more fundamental and powerful than NNs. I like an old statement of DavidFogel, "... True learning cannot occur without evolution. ...". Do you have any feel for EC? Be careful of being trped only within genetic algorithms. That is on a part of EC, albeit the favourite (becasue its so simple any idiot can use it - just like NN backpropagation versus the much [earlier, more powerful, more mathematically correct] ordered derivatives of Paul Werbos 1974,and others even before him). China - I'm very afraid of risks with China, in how they may treat foreign investors, and unseen risks and (incomplete,bious) information are always concerns. Still, some of that alies to Western markets as well. Does the rising importance of the Chinese market all for action with one eye open? Other regions (Europe, India, Russia) - Do you necessarily stick to a USA focus, to avoid spreading yourself too thin? +---+ Weird stuff Human, machine, hybrid - Even with conventional non-(CI, AI machine learning) systems, in my view we've long passed the stage at which human intelligence alone is "economically competitive", and for decades I've felt that I'm really some kind of human-machine hybrid when I'm working (vitual everything I do for my projects is based on computer programming at some stage, usually lots of it). Do you sometimes feel the same? What is your sense of the trend : how much use are human experts in many areas NOW, eg with trading? Here I'm not ignoring a critical role for developing systems. The focus is more on convential expertis, and a growing awarenss from the CI field that essentially all human experts aren't actually very good at their own field of expertise, with rare exceptions (ny guess is <1:10,000, <1:1,000,000 for (creative, revolutionary, breakthrough) thinking). One of my fields of interest is the catastrophic failure of (rational, logical, scientific) reasoning by essentially all scientists, across a areas of science, with a focus on (fundamental theoretical physics, astronomy-cosmology geology, climate, etc). In 2016 I gave a presentation to a consulting engineering firm, with client corporate attendees, on "Big Data, Deep Learning, and Safety". An IBM cognitive scientist corrected me in stating that their (?his?) viewpoint is that not even the exceptional experts (eg medical diagnostics) ca compete with the machines now. Do you feel that way in your area? Fractional order calculus (FOC) - Are you aware of this? Sorry for such an esoteric question, but this whole thing is embarrassing to me. I was shocked & embarassed at my (ignorance, lack of thought) when I came across this in a peer review I did for mathematical theorems for advanced (stability, control) of memristor networks (4th basic electrical circuit element after (rsistor, capacitor, inductor)). Benoit Mandelbrot referred to FOC in his fun book "The misbehiour of markets), and I had intended to look into it more, but didn't. This takes you right back to the war between Gottfried Leibniz and Isaac Newton over who invented calculus. Robert Hecht-Nielson ~2007 "Confabulation Theory" http://www.BillHowell.ca/Neural nets/Neural%20nets/Confabulation%20Theory%20-%20Plausible%20next%20sentence%20survey%20110902%20Howell%20standalone.doc http://www.BillHowell.ca/Neural nets/Neural%20nets/Howell%20110824%20-%20Confabulation%20Theory%20-%20Plausible%20next%20sentence%20survey.pdf Multile Conflicting Hypothesis - Howell, Stephen Puetz MindCode - my #1 project that never seem to get around to work on... http://www.BillHowell.ca/Neural nets/Neural%20nets/Howell%20150225%20-%20MindCode%20Manifesto.odt # enddoc