Market Forecast with Artificial Intelligence

Discussion in 'Trading' started by k3m, May 14, 2012.

  1. k3m



    Relying on state of the art artificial intelligence algorithms, we're starting a (free) market forecast service.

    To pursue our business plan, we need to investigate the needs, qualities and quantities of our potential customers, and since you're all traders here, please take a moment to fill this form:

    It's only 10 short questions! Thank you.

    Note, this is part of a school assignment we have to do for a Business & Economics class. Since we're computer science students, we decided on a very lucrative (provided it is successful) business of providing *reliable* stock and other markets forecasts (obtained with the help of advanced AI algorithms). Please answer the survey as you would if it were a real, soon-to-become available service. You can make four girls and one guy tremendously thankful!

    Note, since it's just a mock business, you needn't provide your email in the last question. :)

    Thank you.

    (Kudos to admins for the forum software. I like it.)
  2. k3m


    So the thread has 100 views, and I only got 6 form submissions. That's kind of sad... :(
    Can you please help a youthful bunch with their homework? :D

    Anybody got any experience with AI playing the markets?
  3. :eek:

    My AI forecast assumed different gender distribution.

    "one guy" – Nice work. Keep doing what you’re doing.
  4. Yes, the market does not like AI, it requires RI.

    RI= Real Intelligence

    Find a project for your class that will offer you a chance to make real money down the road. Don't get involved with extremely complicated matters such as AI.
  5. k3m


    I guess you tweaked it on too few parameters, or perhaps you failed to consider the case of unlikely extreme. On any other day (99% case), you would have been right, though. :) And thanks!

    Well, sorry if that's OCD to you, but I've posted the link to 8 trading forums, and with the amount of views everywhere, I would expect, at 30% click-through rate, at least a satisfiable amount of submissions. Instead, I've now got 13, and I can't really say that's a representative sample.
    I hope you answered our survey. Thanks. :)

    You sure? I'm know AI can compute thousands of weighted tangents per second (e.g. technical analysis) while scraping all the news sources for the relevant news. What exactly is it in the markets, that requires Intelligence? I thought they were openly considered irrational. :)

    And besides, AI is not hard or complicated. Sometimes, it requires only providing valid data and clicking some buttons for the results. :)
  6. Good luck with all this.

    I see you're looking for 80% accuracy.

    That's .. way too high in the real world, imho.
  7. In the future you may run into some classes that explain how advertizing works.

    You also need to consider taking some courses on communications.

    Since AI is the course subject, you may not be able to expect much of your instructor in advertizing or communications.

    To make money trading, AI is unnecessary. This is based upon 53 years of looking at market information. I'm in the 75 to 100 age group.

    Some market information leads the price trading information that is used to place orders. This is real narket information that streams into a computor.

    AI software uses streaming information and puts it through a series of processes to create an AI output.

    The nice thing about what you plan to do is that you can check your performance with other information that actually, in real time, leads the information used to place orders.

    So this means your AI project will give perfect results all the time.

    Since that will be the outcome of all the similar projects in your class, there will be little competition to sell your AI systems. Everyone can have one that they make themselves if they wish to go to the extra work.

    I hope you do several AI projects in your course, the financial industry is not a good application, however.

    See if you can find out why such an assignment would even be made.
  8. k3m


    Well, I guess 80% calls for a definition. :)
    If we were to take accuracy measurements for every single NYSE symbol, then I agree with you. But if we only consider the single symbol with the most probable prediction each day, and count only that towards our accuracy, then I think we can outperform 80%. And that single symbol is all a daytrader needs. But we should provide the whole suite, of course. :)
    Hi Jack,
    We're not advertising just yet. :)
    But what did you mean by achieving perfect results? I didn't understand that part. Can you please explain?
  9. heech


    The survey is flawed, because the assumptions here are broken. This industry (a little like the medical industry) doesn't have typical forms of supply/demand. You can't plot it out and try to figure out the sweet-spot.

    If you actually had a "reliable market prediction" system with a 80% success rate (and equal pay-outs on success/failure), it wouldn't be valued in terms of subscription dollars / month. It would be valued in the range of hundreds of millions, or billions of dollars.

    Going back to my point about the medical industry... it's like going to a pancreatic cancer forum, and asking what the patients would pay for a successful cure. The answer would be: the patients + families would pay every penny they had.

    If you really built the reliable market prediction system you described here, it would be worth every penny in my pocket, and more.
    #10     May 16, 2012