Machines Learn Long Term Fundamental Investing: Part 1
Machine Intelligence has surpassed human intelligence in many fields now:
Checkers: 1990 best human player defeated
Chess: 1997 IBM Deep Blue defeats best human Gary Kasparov
Jeopardy: 2011 IBM Watson defeats greatest human champions
Go: 2016 Google AlphaGo defeats Lee Sedol world Go champion
Naysayer: "But those games don't involve incomplete information like investing does! Machines will never be better than man's intuition in those games!"....
Poker: Jan 2017 Carnegie Mellon Labratus defeats best human players at No Limit Poker
Naysayer: "I got nothin'"
Philosophical Quiz:
1) How does an investor learn and become “better” over time?
2) What does it even mean to be a “better” investor?
Let’s take the second question first since it’s easier – as Warren Buffett says, there are no “called strikes” in investing. You don’t need to have an opinion on every stock, you just need to have ( valid ) confidence in the ones you pick. It ain't what you don't know that gets you into trouble. It's what you know for sure that just ain't so… –Mark Twain. Buffett calls this the “circle of competence” and highlights that the important thing isn’t how large your circle is but rather knowing where the boundary of it is and then only investing inside that circle. The problem for most, especially today, is that the world has changed a lot and many investors are only confident during a certain set of conditions that do not presently exist (e.g. clearly undervalued stocks with good prospects and good management). To summarize a good #2 answer: “The ability to validly define and assess confidence in a wide range of different investment situations.”
Back to the harder question #1, and warning : this will go deep.
Joe v0.0
Let’s start with a v0 model with a hypothetical investor and let’s call him Joe , because Blade Runner 2049... Joe wakes up one day knowing zero about investing and opens today’s newspaper. He picks a company on the front page and looks up the stock: Company XYZ. He researches XYZ intensely – reads all its past SEC financial filings, looks at the historical stock chart, looks through the valuation multiples it trades at today and has traded at in the past, etc. Buffett says “accounting is the language of business” and Joe will learn to read the English language in those filings and conference calls later in v3.0.
Based on this financial data he has now acquired about XYZ, he forms an opinion about its prospect as an investment and decides to buy some shares. Then he waits.
One year later, XYZ stock is up 10%. But S&P is up 12% during that year. While Joe has learned something , it’s clearly not a lot. This has also been a somewhat expensive lesson because Joe actually risked capital in order to learn. Our v0 example is a bit like trying to get good at golf by playing once a year and sadly analogous to how most non-professional investors approach things.
Joe v0.1
Let’s take it a step further. Now v0.1 Joe is a voracious reader. Instead of just looking at one stock XYZ, the day he wakes up with zero knowledge of investing, he looks at every stock that trades in the US (except pink sheets because who even trusts the numbers…). He reads through all the financial numbers in every historical 10-K of every company alive today, looks at every valuation multiple, etc. Then for each of the ~5,000 stocks, Joe decides if he likes it as a long or a short (or neither) and constructs a portfolio of all 5,000 stocks (with some being allocation zero where he has no opinion) based on the data he acquired about each company.
Again, Joe waits a year.
After a year goes by, he has 5,000 outcomes – some stocks up, some down, some bankrupt, some acquired… Joe can start really building a framework of what “works” as an investor – at least with respect to what the world was like over the last year, in an up 12% market… And again it was somewhat expensive because he had to risk some capital and time. How many MBA grads does it take to do 5,000 case studies? How much do they cost?
Part 2 coming soon - for more information please click below...
