Of course....we all know this. Which is why "machine learning" is sometimes a case the dumb teaching the dumber. If you don't have a good market premise to begin with, you're likely just curve-fitting the trades to the price history.
Actually my point was that those who have used AI to crack the unfoldment of time and make exceptional profits are not going to announce it, make a statement or let anyone know about it since: 1) Within limitations it is a licence to print money 2) if it became widely used it would destroy the edge. Those you do hear about are the typical types who go spend $500 on a toy AI software package, take 6 Technical Indicators and their own great idea to measure sentiment throw it into the software and after a few dozen training session with no results declare that AI does not work. This creates something in statistics called Selection Bias. The successes remain very tightly guarded secrets and the failure are complained about widely...so the apparent sample set is 100% failure. Among those declaring the paradigm useless are of course those who have succeeded with it.....an effort to prolong their edge. A few years back there was a full professor in this domain at Harvard who decided to focus in predictive analytics in investing. After a few years of research he was getting noteworthy results and was getting invited to speak at academic conferences in the US and Europe. His approach became more groundbreaking and the results became more exceptional and he was now speaking to a couple of major economics gatherings. Then in the span of a few months he resigned his professorship, his unpublished papers were withdrawn from consideration by major journals and he totally vanished. A couple years later rumors surfaced that he was managing a significant amount of dark money from a private island in the Caribbean.
Cool story if true. Now the "dark money" was from.... 1) drug Lords 2) rich fugitives (like Mark Rich !!) 3) Somalian Pirates Which ??????
I don't have a clue. I read the thing in a post to a newsgroup by a peer of this fellow who was astounded when he and his papers vanished without a word or any logical reason. It could be somebody's wet dream...however it illustrates the likely dynamics of radical upleveling of technology that manufactures money. Unlike the new IPhone which requires manufacturing plants, ad agencies, media buzz and a million retail sales to make billions, cracking the egg of time in investing calls for the reverse: secrecy, subterfuge and secret agent tactics. Something that can make billions with only a thin and tenuous channel to execute trades quietly and fit on a thumb drive is a new kind of economic animal.
Just relaxing in da sun man Da big black spider reached out for 'im He work all day while we play. :eek:
Microsoft releases open source artificial intelligence toolkit CNTK on GitHub A toolkit "insanely more efficient than anything we have ever seen" http://www.winbeta.org/news/microso...rtificial-intelligence-toolkit-cntk-on-github
Anyone who evaluates this please post your observations. Initial impression is that the focus is image and Voice modeling. Unless there is something I don't know (often happens) this might be a stretch for market models..... Hmmm...wait on second.... a thought .....treat the market as a voice that is saying up, down or unknown...??? It might well be great for cycle decomposition and identification.
Whilst MS and Windows have their quirks not sure why the sarcasm re Microsoft and AI? C Bishop who is quite well known in the ML field works there and so do other well-regarded ML researchers.
I missed the other obligatory buzzwords of our time: "cloud", "cloud computing", "big data". Hmm. maybe a change at MS? LOL How many APIs has MS brought, which only some years later then were revamped? I guess more than 50. The poor programmers who had to learn their useless crap.