Unknown and Unknowable
Richard Zeckhauser's 'Unknown and Unknowable' and a Theory of AI's Investing Application
Building on Richard Zeckhauser's essay "Investing in the Unknown and Unknowable" I present an application for AI in an investing context. Under certain circumstances, AI is less useful in predicting hard-to-predict outcomes, but extremely useful in identifying when the market is failing in the face of unknown future states.
This view of AI does not rest on a belief in its superior analytical ability, rather on its ability to continuously evaluate data relationships and identify when they have broken meaningfully from rational or historic ranges1.
“Discounting for ambiguity is a natural tendency that should be overcome, just as should be overeating.”
In 1996, a consultant for the California Earthquake Authority was looking for $1 billion of re-insurance that would take effect after $5 billion in aggregate insured earthquake losses. Traditional financial buyers would not bite even after the authority offered rates five times the actuarial value. The odds were just too unknown using even the most advanced statistical models. As told in Richard Zeckhauser's "Investing in the Unknown and Unknowable" Warren Buffett flew to California and took all $1 billion of exposure on to Berkshire's books.
Thus, Buffett—despite attention to money management—was willing to take 100% of a risk of which Wall Street firms rejected taking even part. Those fancy financial entities were not well equipped to take a risk on something that was hard for them to estimate. Perhaps they did not recognize that others had no inside information, that everyone was operating with the same probability.
Such is the key theme of Zeckhauser's essay: Some of the best investing outcomes result not from greater skill in discounting, but rather in recognizing a lack of any attainable edge. In such situations, asset prices often overly discount ambiguity.
The real world of investing often ratchets the level of nonknowledge into still another dimension, where even the identity and nature of possible future states are not known. This is the world of ignorance. In it, there is no way that one can sensibly assign probabilities to the unknown states of the world. Just as traditional finance theory hits the wall when it encounters uncertainty, modern decision theory hits the wall when addressing the world of ignorance.
Zeckhauser wrote long before the technological potential of Artificial Intelligence, Machine and Deep Learning were known. In his essay he outlines nine "maxims" for investing in Unknown and Unknowable ("UU") situations. I believe these maxims, at least in part, lay out a framework for how artificial intelligence may ultimately be applied to the problem of investing. The bottom right of the below table is the purview of artificial intelligence and machine learning.
Zeckhauser's Nine Maxims Paraphrased
Individuals with complementary skills enjoy great positive excess returns from UU investments. Make sidecar investments alongside them when given the opportunity.
The greater your expected return and the deeper your advantage, the greater the percentage of capital you should put at risk.
When information asymmetry leads a counterparty to be concerned about trading with you, identify your absolute advantage.
In situations where probabilities are hard for either side to assess, it may be sufficient to assess your knowledge relative to the other party or market.
If your advantage is significant enough you can invest even if the outcome is UU
Even sophisticated investors tend to underweight how much the value of assets can vary.
Be triply beware of herding when there is evidence that there have been significant changes in the basic structure of markets, however stable they have been in the past.
Discounting for ambiguity is a natural tendency that should be overcome.
Do not assume that just because you do not know the risk that others do.
Identifying UU Situations
Zeckhauser illustrates through multiple case studies that UU situations create unique opportunities for generating positive excess returns, but he also identifies that profiting is more a matter of assessing the likely edge of your counterparties than accurately handicapping the future. Zeckhauser pulls an example from the musical "Guys and Dolls",
The overseer of the roving craps game is Nathan Detroit. He is seeking action and asks Sky Masterson—whose good looks and gambling success befit his name— to bet on yesterday’s cake sales at Lindy’s, a famed local deli. Sky declines and recounts a story to Nathan:
On the day when I left home to make my way in the world, my daddy took me to one side. “Son,” my daddy says to me, “I am sorry I am not able to bankroll you to a large start, but not having the necessary lettuce to get you rolling, instead I’m going to stake you to some very valuable advice. One of these days in your travels, a guy is going to show you a brand-new deck of cards on which the seal is not yet broken. Then this guy is going to offer to bet you that he can make the jack of spades jump out of this brand-new deck of cards and squirt cider in your ear. But, son, do not accept this bet, because as sure as you stand there, you’re going to wind up with an ear full of cider.
Keep reading with a 7-day free trial
Subscribe to Lewis Enterprises to keep reading this post and get 7 days of free access to the full post archives.