Greetings --
My posting was intended to show the extent to which artificial intelligence and machine learning is being used by large trading houses. I am surprised by the messages posted in response. They cover a wide range, and many imply denial. Let me make a few comments.
None of my comments are intended to be critical of individuals or their trading and investing techniques. If something works for you, ignore me and keep it up.
1. Some comments suggest that fundamental analysis is more important than quantitative / technical / statistical analysis.
In the 1990s I was hired by a company based in a southern US state. They wanted a stock selection and signal generation program that could be sold to individual investors / traders that used fundamental data for the stock selection portion. I was vice-president for research and in charge of the project. Years earlier -- late 1960s and early 1970s -- one of my projects as a graduate student had been a stock selection computer program, based on artificial intelligence, that worked. So I thought we might succeed. We tried everything we could think of, including analysis of a variety of corporate fillings. It was during this period that my analysis of the use of fundamental data crystallized. A recent version of that has been published several times, and is an appendix to the "Foundations" book. The paper can be downloaded from the book's website:
http://www.blueowlpress.com/wp-content/uploads/2016/10/FT-Fundamental-Analysis-Appendix-A.pdf
During that period, a professional colleague was working for a highly respected research organization that published (at a very high subscription rate) buy / hold/ sell recommendations. His mother lived in a retirement home in my community. On his regular visits to her, he stopped in my office and we talked about my project. His summary when I asked him "What works?" was "Nothing we have found. That is, nothing outside the board room."
My group could not find anything either, and the stock selection portion of that project was dropped.
2. The world's richest person uses buy and hold. I am assuming the reference is to Warren Buffett.
Mr. Buffett is as much the president of a conglomerate corporation as he is an investor. By his own assessment, he is not a trader. The price of the Buffett-led Berkshire Hathaway stocks (BRK-A and BRK-B) dropped over 50% from early 2008 to the lows in 2009. In my opinion, holding resources through losses that large is more a characteristic of a corporation than of an investor or a trader.
The 50% drawdown was recovered. In my opinion, the recovery in equities, bonds, and real estate from that "crisis" has been the result of low interest rates set by central banks. There is a very high risk that as interest rates rise, the majority of those gains in all three areas will be lost. We are already seeing some of the effect of rising interest rates on bonds.
3. Productivity gains will improve company profits and raise all boats.
Several political economists have pointed out that the "developed" countries benefited from recovery following world war II, which provided strong growth and rising values of all investments beginning about 1945. That increasing globalization has recently been spreading the effects throughout the world, and that we (the US, at least) should not expect strong growth in the 21st century.
Some of the gains in productivity are due to automation. Those profits accrue to stock holders, but not to workers. Investors may benefit, inequality may increase.
4. Some comments suggest that AI/ML is unnecessary -- that simple techniques work.
I was a senior research analyst at a commodity trading company (a hedge fund) in the 1990s. The company had been one of the pioneers in development and use of trend-following quantitative systems. Their specialty was breakouts -- buy new highs and other "turtle-like" techniques. We watched, literally day by day, as those techniques stopped working. In my opinion, they will never work again. Trends are shorter in duration and price movement. Systems that take counter-trend positions are quicker to fade the trends, breaking them into smaller pieces which are less easily identified and less profitable. That company returned all investor money and ceased operation.
But assume that some simple indicator does work today. Take Stephen Simmons comment about J. P. Morgan employing thousands of individuals to develop trading systems as literal and think through the implications. We know from many articles that research personnel are highly educated -- one-third or more have PhD degrees in scientific disciplines -- and highly compensated -- say $100,000 base salary on average. 1000 people at $100,000 per year is an annual salary cost of $100,000,000. Add office space, support staff, benefits, etc and the sum grows. Morgan will not build those departments and pay those costs unless the trading results are adequate to fund them.
A trading company does not need 1000 people and a department budget of several hundred million dollars to tell them to buy when the 5 day moving average rises up through the 20 day moving average.
Anyway -- thanks for listening,
Best regards, Howard