Morgan tweeted a fun question today
What is the most powerful and consistent law of investing?
— Morgan Housel (@morganhousel) December 1, 2020
It’s a provocative exercise to reduce investing to a single law. There are many thoughtful responses. It’s a healthy opportunity to consolidate your core beliefs about investing.
Twitter has exposed me to many ideas that I’ve re-mixed with my own experience. That process may uncover universal truths. It has helped form new beliefs, challenged old beliefs, and improved my understanding of some beliefs. I maintain a notebook of these beliefs.
There are 3 categories:
- Current Priors
Examples: Long term bonds offer poor risk/reward or demographics are slow moving drivers of growth rates
- Beliefs Graveyard
Examples: Profit margins should mean revert, printing money leads to inflation, more risk equals more reward
- Evergreen Beliefs
I have publically shared these on my wiki.
I have recently added another to my “current priors”. It’s a belief that has underpinned my trading style for 20 years but I was only able to name it recently. Why? Because it only revealed itself in relief to what other people think investing is about. By observing how people think about investing, I’ve noticed my current prior is not widely obvious:
Investment edge is about measurement not prediction.
The distinction between prediction and measurement
Predicting means you make a probability-weighted opinion of future states of the world. This is typically built up from logic and reasoning.
Measurement is counting. Usually with an intent to rank. The idea is simple but not easy. Accounting is an entire field devoted to trying to create a meaningful picture of a business by normalizing quantities across industries.
These are of course not mutually exclusive activities. Measurement is done in current time and then projection leads to predictions. The prediction eventually becomes a point in the past, gets counted, and iteratively fed into the next projection.
Yet when I say edge is about measurement not prediction I am very much emphasizing the importance of measurement over prediction. There is just a single prediction underlying the entire framework:
Over time, cheap stuff outperforms expensive stuff.
But there is so much to measure. Once you accept this, your gaze shifts from the future to the present. The focus becomes finding the variables that matter, harvesting and cleaning data, and normalizing across markets. Every one of these steps leaves tremendous opportunity for creativity, discretion, and disagreement.
Why do I make this distinction?
There are many ways to skin a cat (don’t linger on that too long, it’s an expression better left unexamined).
Your cornerstone beliefs depend heavily on the path your career has taken. There are activist investors who use their rights to influence outcomes. Thematic investors concentrate their portfolio around a vision of the future (AI, crypto, cannabis, clean energy). They are the epitome of crystal ball predictors wagering their money and time. These strategies are high risk-reward with long feedback loops.
But I came into the investing world from the nearly the opposite angle. The market-making side. I see trading as the role of the house in a casino. That means diversifying over a wide range of discrete edges and not letting any single bet risk putting you out of business. Rapid feedback loops and large sample sizes. The focus is on operationalizing and measurement. In options, if you do not understand volatility time you cannot compare implied vols across markets. There’s only one major prediction: 1 If I measure cheap and expensive correctly it will pay off as surely as the casino gets paid.
If the distinction isn’t clear consider this story from Andrew Tsai about his internship at Susquehanna (disclosure: I worked there for 8 years after college). He recounts a company outing to a dog track:
I’m sitting next to one of the partners and I’m looking at the sheet of all the races, and he’s like “How are you gonna bet?” I respond, “Well, I’ve never really done this before but this dog looks like he’s got a good track record and he’s been running strong lately.”
The guy looked at me like I was a complete idiot.
He’s like, “What are you talking about, ‘How is this dog doing?’”
Andrew is perplexed. Well, isn’t that kind of what we’re talking about.
The partner starts to explain, Look at the relative value of this dog and that dog.
Tsai’s confusion was at the heart of prediction vs measurement. The partner didn’t have any opinion about the dog’s chances. He was just looking for mispriced lines. Which is another way of saying, “I don’t know what is going to happen but these prices are self-contradictory compared to the future state of the world.”
Yes, there’s some embedded prediction there, but the framing is very different from wondering which dog is feeling extra feisty.
Measurement — simple but not easy.
If you normalize correctly then the cheap and expensive are just colors on a heatmap. In options, you must normalize vols, time, funding, and any parameter you believe drives pricing. Your feedback loops are punctuated by expiration cycles where the asset’s distribution needs to be settled. That’s the beauty of options. My friend Jeff explains this beautifully with respect to futures markets in his interview with Corey Hoffstein. Expirations force a convergence of truth. Manage risk, learn, repeat.
Exporting the logic to fundamental investing is outside my practical experience, but I can see the complexity of the problem. Consider stocks. As perpetual claims on future earnings, they have durations that vary with market discount factors. Today, fundamental investors feel like the ref (aka discount factor & interest rates) has tacked 90 minutes of extra time to the soccer match as earnings are amortized from ever further into the future.
I’m even more daunted when I consider wrangling fundamental parameters. A superficial process doesn’t distinguish between a value stock and a value trap. Even deeper, there’s a big difference between a stock whose cheapness reflects a low forward ROIC vs a stock that’s cheap because the market knows its inventory is overvalued.
From the outside, it appears that quant approaches to stock-picking come from the measuring camp. Yet commoditized approaches are likely frustrated by naïve measurement practices or practices that do not truly scale. So long as reading the fine-print in financial statements is labor-intensive and dependent on judgement, stock-picking will be an opportunity for analysts at the pinnacle of the art of measuring.
In addition to understanding business, there’s an adjacent skillset of understanding finance. In this thread by @10kdiver, we can see how stock based compensation can obscure the cap structure of a company. Any valuation estimate downstream from this analysis will be sensitive to how this impact is normalized across stocks’ dilution paths.
Selling a differentiated depth of knowledge in both finance and industry is easier than actually possessing it. So it makes sense to have a high bar when faced with someone claiming to have both.
A focus on measurement and agnosticm
I’m probably one of the few individuals who has ever leased on a seat on the Amex, NYSE, NYMEX, COMEX, and NYBOT/ICE (so every exchange in NYC. You are welcome to view this as a demerit as well).
By the end of my floor trading life, I leased (believe me I wish I owned) seats on all 3 exchanges in the NYMEX building at 1 North End Avenue. I was a ghost who would migrate from pit to pit based on where the action was. I learned the specific dynamics of several markets including oil, gas, precious metals, and soft commodities. Yet, I only dove into fundamentals deep enough to understand the major “gotchas” (usually squeeze mechanisms and bottlenecks) and seasonality trends. The Slovic study on horse bettors who were given more info than they needed showed that the extra data increased their confidence faster than their accuracy. And as you’ve heard, “It’s not what you don’t know that kills you, it’s what you know for sure that ain’t true.”
Find the variables that matter and ruthlessly tune. Normalization within and across these markets became the focus. I was a microcosm of the firms I had worked for. The specifics of a market needed to be learned to be fed up into a more bird’s eye view. Then it was a matter of sorting cheap and expensive. At no point did I have an opinion on underlying price. I’m a donkey. I just look for mispriced point spreads. Yes context matters, but as much as possible, I held an end goal of feeding normalized parameters into a framework to see what pops out.
Today, I remain skeptical of thematic and fundamental approaches. The former because it relies on prediction. For fundamental approaches, I’d like to see how a manager shortens the feedback loops. How they can map the variables they focus on to outcomes. Ideally variables that bounce around so there is trading opportunity and sample sizes. If this process lacks iteration it cannot improve. It just starts to look like thesis investing.
My bias (again, views are path-dependent and I am clearly giving my experience lots of weight) is towards agnostic, opportunist type of investors. Folks that are normalizing lines rigorously to find the mispriced horse. No allegiance to asset class or sector. When I talk to a manager I want to know what kind of metrics they look at, why those metrics matter, how they know if they still matter, and what short cycles they study to inform the metastability of their framework.
You are either a casino or a tourist. So if you claim you can count cards I need to be convinced they are countable. Can you measure?