How much are you hurting your returns?
Robotic-Systematic or Systematic-with-a-Human-Twist?
Before we get started, I note that this article’s title was totally stolen from Ian Cassel’s (see below) article. Sorry Ian, it was just too good.
Searching for value on Twitter
As much as I hesitate to admit this, I have a Twitter1 account. Despite Twitter being predominantly a rancid wasteland of p0rnbots and scammy advertising, occasionally, very rarely, someone I follow will post a gem and I can tell myself wading through the Twitter muck was worthwhile.
Recently, Ian Cassel of The Microcap Club posted a link to just such a gem. In his referenced blog post, How Much are You Hurting Your Returns Ian describes an interview with the CEO of Norway’s sovereign wealth fund. The fund is massive… they apparently own the shares of approximately 12 000 companies, comprising roughly 1.5% of all the equity value IN THE WORLD.
As Ian notes (and I agree), perhaps the most profound insight from that interview was the question of how to judge whether your own input into your investment process is adding, or detracting, from your performance over time.
"The way to judge yourself is inertia analysis. Run your January 1 portfolio for the full year without any changes and compare it to your actual results. It's awful because some years you realize all you did was subtract value when you went into the office."
And this is relevant how?
I’m a systematic investor. My failed earlier attempts with other investing styles left me with a proven track record: I have zero ability to forecast macroeconomics, no ability to detect lies from bad CEOs, can’t correctly pick “turn-around” companies, and I have not a clue which industry will do well in the next 6-12 months, let alone which individual company’s shares might shoot up suddenly for a quick profit.
In the face of of my demonstrated macro and behavioral investing deficiencies, and perhaps due in no small part to my programming and mathematical background, what I developed instead was a system and software to support a systematic, rules-based investment process. Programmatically sift through a universe of stocks, discard the obvious turds, rank the remaining candidates in terms of desirability, allocate available funds to the best identified investment opportunities at that time, hold until the facts change. Easy right?
Well sorta…
You see, I do not invest robotically.
Robotic Investing
Robotic investing represents the most extreme version of a systematic process — come up with a set of rules dictating what to invest in and how much money to invest, and follow that process strictly, absolutely, and with zero human intervention. Never override the process, never peak inside the machine to check what it is doing, simply follow obediently (some might say, blindly) at all times.
I don’t do that.
Its true — I developed myself a set of rules to find great stocks and tested a set of proven approaches to hopefully guide me through volatile times. But I know enough now to say that between quirks and bugs in financial market data, bizarre-o market interventions made by bankers and politicians in times of crisis, and the simple fact that we live in a messy world that is constantly changing, a purely 100% strictly robotic approach is unlikely to ensure success. I believe there will always be a need to cast an eye over the TCS rules-based, software-generated stock rankings, looking for oddities and then performing research to review and subsequently approve/reject the stock picks before committing to buy or sell.
When robotic rule-following fails us
As a simple example, imagine we aim to achieve a 15% return annually for our portfolio. We might have backtested a rule that said “After buying shares in a company, if the share price has not risen at least 15% annually after 3 years, sell that company regardless of how the business is performing and invest that money elsewhere”. The reasoning behind such a rule might be that you buy what looks like a good company according to your selection criteria and then you allow enough time for the business quality to be reflected in the share price performance. However, after holding for an extended period of time, for some reason “The Market”2 doesn’t agree and the share price isn’t increasing as much as your target rate. Perhaps there is some quality or characteristic of the business that your model doesn’t capture, so what looked great in theory is for some reason not great in practice. Your rule says that you should exit that stock and try something else. You trial this simple rule in your backtests, and over the long run it appears to improve your expected results.
Most of the time, this might actually a pretty good rule. During backtesting it triggers sales of laggard stocks and rotates your money into new, fresh opportunities, but not so often that you’d be trading excessively and paying increased fees and taxes. So far so good.
The problem is… something like COVID comes along in early 2020 and EVERY stock slumps in price. Suddenly businesses with incredible profits and performance like Facebook and Applied Materials are available at the same price they were back in early 2017! A naive, robotic implementation of your rule triggers because “AMAT stock hasn’t risen in 3 years, it must be bad for some reason, sell and move on”. Robotic, mindless following of the rules would have you selling out of amazing businesses, right when you should probably be buying as much as you can to take advantage of discount prices!
At such times, robotic rule-following would fail us, and so I believe the model should be overriden. The momentum of our rules rolls our investing car in generally the right direction, but we keep a human hand on the steering wheel.
Evaluating MY input
So if I don’t mindlessly follow the rankings and recommendations of my investing software, and I do at times use some human discretion in my stock selection process, how then will we know if my guiding hand is a net positive or a net negative for the TCSI portfolio results?
Behind the scenes, in addition to the systematic-but-not-robotic portfolios I release each month, I have been and will be recording a parallel set of portfolios. A pure, no-touch, 100% robotic portfolio that strictly adheres to the TCSI stock rankings and buy/sell recommendations — No Humans Allowed.
In this way, I heed the advice of the Norway Sovereign Fund in a very practical, analytical manner. I can compare in real time whether I add or subtract value every time I walk into the office. In the long run and depending on the outcome, I expect these parallel portfolios to either reinforce my belief in the benefit of an occasional human guiding hand, or alternately, with enough proof, suggest that I actually step back even further from the controls and let the rules do all the driving.
I’m equal parts nervous and excited to see if, with the right set of rules and through a range of market environments, I can make myself redundant. Please subscribe to follow along and find out.
Yeah Twitter, not “X”. Never X.
Read: All the other investors out there