Awful Augurs
Vol #84: June 3rd, 2022
The most enduring parlor game of mankind is the attempt to predict the future.
We poke fun at the weatherman who can’t tell you to bring an umbrella two days ahead, or the fortune teller that contrives love stories from the depth of your palm creases. Yet every day we make decisions that are based on some estimation of the future state of the world.
At the extreme end of certainty we predict that the sun will rise at about the same time as it did yesterday, and that taxes will be due on the money that we’ve earned. Wading further out, we plan a ski trip in February because there’s a good chance that there’s snow in the Rockies then. We put our long term savings in the stock market, because we’ve been told that across most investment horizons the value of our nest egg should go up.
For big ticket items, our estimates of the future tend to be rough and broad. The underlying signal tends to be strong, and the long horizon dampens the short term noise. The shorter the duration, the more randomness comes into play, and the more novel the context the less certain we can be.
Sports fans make an implicit prediction about the future of the game. By investing time into following players, watching matches, and organizing Sunday social activities around games, they are betting that this activity will continue to be a source of joy. We can give a high degree of confidence to the NFL being around in 10 years, but only a Las Vegas sportsbook has any confidence in the outcome of the next game.
Statisticians have turned sports gambling into a fine science with mountains of data, but as the propositions get more arcane, confidence decreases exponentially. Online betting (public liquid markets) increases the efficiency of pricing, and the lines on any given match are incredibly accurate. Who will win the Giants / Patriots game is a very fairly priced bet to make, but how many yards the quarterback will throw for has a lot more uncertainty around it.
Financial markets are fundamentally a mechanism to invest in, speculate on, and hedge predictions for the future. The prices on the tape tell the story of how much weight the market gives to any single potential outcome.
Just like with the sun rising, the best prediction of tomorrow is that it will look a lot like today. It’s the closest data point we have to the future uncertainty, and the seismic shifts described in history books are but a fraction of the lived experience. (According to clock time at least, there’s a very interesting argument that emotional time is not linear and that the gravity of significant events warps our perception of the past.)
Options are derivative instruments, and thus their value is directly related to the underlying security. At settlement their value is purely a function of the terminal value of that underlying. But at any given moment before that settlement, the options pricing model assumes that the settlement will be at the current price, with some layer of extra value depending on how much time is left and how volatile the stock is.
Predicting volatility then becomes a next order of complexity in trying to make a financial estimation of the future. Again, one of the common techniques is to simply look at how things are behaving currently, and leaning on those assumptions. When a new options contract month gets listed, the default value for that volatility estimate is an interpolation of what surrounding months are pricing - perhaps with some nudge up or down based on a known event like earnings.
Markets can also provide us with a distinct probability of a given event happening. When Elon Musk says he is going to buy Twitter for $54.20 a share by the end of December, the current market price is an estimation of the probability of that deal getting done. Subtracting the cost of capital, the ratio of the current price to the price before the deal gives an indication of how likely it is to close. If Twitter was worth $35 a share before Elon opened his mouth and checkbook, with it trading around $39 today means the market only sees a ($39-$35)/($54.20-$35) or ~21% chance of this deal getting done at that price.
There are a lot of ifs that go into that prediction. Being able to enumerate them has the potential to seriously cloud our perception of the same. The quantum fog of long term uncertainty is easier to judge objectively.
Experience and historical data can help make better predictions. Knowing the Patriots have won their last 2 games against the Giants and are up 7-6 lifetime in the series is a good baseline assumption to start with. That’s not all that dissimilar from saying that tech stocks tend to outperform the broader market.
For the ten years going into the close of 2021, the NASDAQ index is up 560% compared to the S&P 500’s gain of “only” 266%. But year to date in 2022, the NASDAQ is down 22% compared to 13% for the S&P 500. (Combining these horizons, you’re still a big winner with the NASDAQ; 447% vs. 226%)
The main narrative here is that the fundamental assumptions of the prediction model have changed. If stocks are the value of the future revenue of a company, when stocks fall the estimations of future revenue have also dropped off. It doesn’t take that much of an interest rate change to dramatically impact the long term discounted cash flow of a speculative growth company that relies on a low cost of capital.
Traditional markets have a long history of company valuation. We’ve been through many cycles of changing macroeconomic conditions and observed their impact on the business prospects of different companies.
Vanguards of technology like to believe they’re tearing up the playbook, but so long as they are grounded by the gravity of the New York Stock Exchange and Sarbanes Oxley, the CAPM pricing model and institutional allocators will be the arbiters of their future. When market rates go up, heady growth assumptions need to change, no matter how dazzling the engineering is.
But how do we predict and value something that has only a short history and a mercenary’s rule book?
Digital assets fluctuate wildly because they are unbounded zeitgeist with unlimited potential outcomes. Tokens can magically appear in a user's wallet via airdrops for using a protocol. If there’s a chance that buying ice cream from Mr. Shane’s Homemade Parlor Ice Cream will result in getting a piece of Mr. Shane’s revenue, how much should your cone cost? That’s a little different than predicting Exxon’s next dividend.
Prediction is our favorite game to be bad at. Even within a well defined landscape, knowing when a zig becomes a zag is nearly impossible. The constantly shifting landscape combined with human biases makes estimations about specifics very difficult. As we push out the timeline or reduce the specificity, broad trends emerge. It’s usually more of the same, until it’s not.