You have probably heard about three AI machines that are readily available. Google has its Google Home. Amazon has Echo. And Apple has Siri. I own all three just so I can watch how they evolve. I’m told that in Silicon Valley now, almost all new startups seem to involve some form of AI. AI is projected to be a major growth area for the economy in the coming years. In fact, if you want a fairly safe portfolio during this bull quiet market, you might simply consider owning Apple, Alphabet (Google), Amazon, and Nvidia (which makes the chips for most of the AI computing). I personally own three of those four in my retirement account.
AI is getting to be very useful for everyday purposes. When I was riding in a private cab from Heathrow to downtown London last month, the driver used an English smartphone app to navigate. It was more sophisticated than any driving assistants that I’ve ever heard in that it used the same voice to process directions and pronounced street names very clearly. Periodically, it would also say something like, “There is heavy traffic on ABC road but I can save you 17 minutes if you turn left on XYZ road ahead.” To me, it actually sounded like a human being somewhere was monitoring the traffic on our route and giving advice on how to proceed. And it was just a UK smartphone app.
I mentioned my interest in AI companies at our London systems workshop and one of the attendees asked me if I personally knew of any AI machines that were trading. I mentioned that there are probably trillions of dollars being traded through computers, but I also said that kind of computing (AI) works best for closed systems. For example, IBM’s Deep Blue beat chess grand master Gary Kasperov in 1997. Chess is a closed system where the rules don’t change. The trading/investing arena is anything but a closed system because the rules change constantly so I have been skeptical about what was possible with AI and trading.
To my surprise, two days after my return to the United States I found an article in the Microsoft Money section of my computer on a new ETF that was being run by IBM’s Watson computer.
In case you were not aware, IBM isn’t a giant hardware manufacturing company anymore. It uses its massive computers as AI consultants in many, many areas. You ask it questions and it responds. Watson is an AI computing platform named after the founder of IBM, Thomas J. Watson. In 2012, Watson beat two of the best champions of the hit TV show, Jeopardy, to win a million dollar prize. Obviously, Watson is very good at answering questions and I guess the Jeopardy appearance was a good contract for IBM.
IBM’s Watson received lots of publicity from its Jeopardy appearance and now there lots of companies are using it. For example, Watson has been hired by the Wimbledon tournament to develop highlights based upon things such as crowd noise. GM uses Watson to develop AI in its cars. In addition companies are using Watson to 1) develop advertising and marketing strategies, 2) develop social media marketing, 3) interact with customers, 4) develop tasty new recipes, 5) manage patients, 6) make decisions about buying new capital equipment, 7) improve the banking experience with customers, and 8) help in buying a home. The list goes on and on and Watson is even involved in 9) weather forecasting.
A San Francisco firm, EquBot, hired Watson to run a new ETF (symbol AIEQ). AIEQ, launched in mid-October this year, picks stocks using over a million pieces of information available each day to update its assessment of the over 6,000 publicly traded companies. It then picks 30 to 70 stocks to own based upon their probability of benefiting under current economic conditions.
It looks like EquBot uses Watson to mimic an army of equity analysts working 24/7 — but removing human error. That approach, however, misses all of the errors included in the pre-suppositions of such an investing strategy. Right now, the ETF is an automated version of investing with all of its human biases (such as an emphasis on stock picking).
That’s my main concern with an AI investing strategy — it includes all the human biases about what’s supposed to be important to good investing. In my opinion, most of those “important” elements are actually irrelevant and they massively compound the complexity of investment decisions. Even though the AIEQ strategy continues learning, understanding how little is truly important for successful investing would be a lot for a machine to unlearn as it moves along in its journey. For example, could a machine actually learn the importance of position sizing strategies without being biased in that direction?
Here are some of the questions that I have thought about after I heard about AIEQ —
- Does it understand reward to risk ratios in its decisions?
- Does it know when it’s wrong about a position and then exit? Or instead, does it close a position when it finds a better trade or when it thinks the information about the current position a top candidate is no longer useful?
- Does it understand the power of exits and the value of “not being in the market” under certain conditions?
- What will its performance be like during huge down markets?
- What will its performance be like during sideways markets? Or any of the various market types?
- Is it biased toward asset allocation? (Because very few humans really understand what’s important about asset allocation). For example, if it has a bias toward some sort of asset allocation built into its programming, it might be very hard to overcome that bias.
- Does it know the power of position sizing strategies?
- Is it biased toward being 100% invested at all time? If so, can it go short when the market is going down?
Anyway, I thought the idea of an AI fund was interesting enough to risk a small 0.5% equity position for AIEQ in my retirement account. That will help me watch it closely. I have also realized that IBM has become a not-to-be-ignored AI company now.
In a few months, we will add AIEQ to our World Market Model database of about 500 ETFs that we monitor for the monthly SQN Report. We’ll need 100 days of price data first, however, to generate its Market SQN score.
Since hearing about AIEQ, I have started to think about what Watson could do if it understood Tharp Think and it managed an ETF with those principles. If Watson is sophisticated enough and could understand things like market type and the interaction of different systems, then this kind of AI based ETF could actually work well. Perhaps that should be a new business venture for me — or would that be taking me too far off purpose? It’s just food for thought right now.