There are a number of firms out there that are trying to marry Artificial Intelligence (AI) and Investing. Machine Learning is closely related to AI, but slightly different, insofar as you need the AI before the machine can learn anything. The goal is to “code (computers) to think like human beings, and then plug them into the internet to give them access to all of the information of the world (Forbes Magazine, 12/6/16).” There is so much Past Performance information out there on all types of securities that the hope is that, through AI, computers can learn and discern patterns from all of this information and can make wise investment decisions parsed from all of that data.
I am not computer savvy-enough to develop my own AI/Machine Learning program for trading securities. However, I did for a while employ an outside service that had developed an excellent machine learning program to predict market downturns in certain specific securities. There are a lot of services that try to do this but I believe the service that I used was the best out there. Without giving too much away, the service would send me an email whenever it calculated that there was about to be a market downturn. It was up to me to determine what to do when I got that email. I found that the service was accurate most of the time. However, the overriding issue during the period that I used the service was that the market was in an overall upward trend. Any “warning” signals that my AI provider gave me turned out to be short-term. The service did not follow up with an “all clear” email once the downturn was over, although the service has since developed something similar that works on the long side as well as the short side. I have not used this service since we have had market turbulence dating back to February 1, 2018.
Instead of the AI service, I rely on overall market indicators. While these may not be as up-to-the-second as my AI service was, I find that they are as accurate in discerning the large trends in the overall market. My indicators tell me that we are still in a general uptrend but there will continue to be hiccups. We won’t go straight up again but up is the overall general trend. As a result, I am finding that I am holding on to positions for a longer period and not getting caught up as much in day-to-day market fluctuations.
Having spoken with other investment managers that have used or are currently using AI/Machine Learning tools, I find that my experience is not out of the norm. These systems are nice to have but don’t bet the farm on them. They work if the data that they crunch plays out in a manner that is consistent with the past, and they are useful if they are correct more than half of the time or if they show that their trades are more profitable than they are not. I wrote in my previous blog about the necessity of continual reinvention. AI makes such reinvention more possible, but there is never an end to it. For the individual investor, it is better to stick to your plan of investing in a diversified basket of financial assets and leave AI to the rocket scientists. AI investing may one day have mass appeal but it is better right now to invest the old-fashioned way, through your own diligence with help from investment professionals you trust.