Decisions Under Uncertainty

How luck, skill and hidden information play a role in successful investing

To begin our discussion on decision making, we can try to imagine we are at a casino, looking around a vast number of games being played.

By analyzing these games through the lenses of luck, skill, and the presence of hidden information (i.e. what is known and unknown), we can create a helpful structure for understanding how each of these elements contributes to winning or losing.

Poker for example, involves a combination of luck, skill, and hidden information. In contrast, chess is primarily a game of skill, while backgammon involves both skill and luck but no hidden information. At the other end of the spectrum, Russian roulette is a game entirely reliant on luck. Furthermore, the definition of ‘skill’ can be difficult to pinpoint. If skill involves making intelligent decisions in the face of uncertain outcomes, it could also be argued that assessing and managing this uncertainty is a key component of skill.

One key point from Annie Duke’s popular book “Thinking in Bets: Making Smarter Decisions When You Don’t Have All the Facts” is that “good decisions” do not necessarily produce good outcomes, or in other words, the quality of a decision cannot be judged by their outcome. Just because you scored a “hole in one” does not necessarily mean that you had a good practice routine or a well-rehearsed swing - it could just be a matter of luck. It is important to separate decisions and outcomes, as outcomes - which society places a lot of weight on - can be influenced not only by skill but also by luck or by the presence of hidden information that was unknown at the time the decision was made. As a result, decisions that were thought to be “good” may fail, while those that were given little thought may succeed.

If we now leave the casino and consider the desk of an investment manager, we can see how their ‘game’ resembles poker more than chess. As Howard Marks notes in his memo discussing Annie’s book, the three attributes used to analyze games also apply to the world of investing. Firstly, there is hidden information. No one (even those inside the company) has access to all the information. Companies disclose financials and forecasts, and investors meet with companies to gain more insights, but the fact is that only a portion of all information is made public and taken into account. This contradicts Eugene Fama’s “efficient market hypothesis,” which states that markets are “efficient,” all information is taken into account, and there is no room for “skill.” Investors reject the hard form of this theory, and the crux of the debate is really about how “efficient” different markets are to varying degrees.

Secondly, there is element of luck: unexpected events such as wars and pandemics can have a significant impact on the performance of our portfolio and occur completely randomly. When they do happen, they often have a disproportionate effect on the performance of our investments, affecting both skilled and unskilled investors equally.

Thirdly, skill: good investors are said to have the skills to better assess the future cash flows of companies, navigate industry changes, different geographies, and market cycles. Many investors claim that they are able to generate “alpha” (or a return above the average market rate) thanks to these skills, which they believe deserve to be compensated. Annie argues that what good poker players and good decision-makers, such as good investors, have in common is their comfort with the uncertain and unpredictable nature of the world.

“They understand that they can almost never know exactly how something will turn out. They embrace uncertainty and, instead of focusing on being sure, they figure out how unsure they are, making their best guess at the chances that different outcomes will occur.”

This method of ‘best guessing’ or quantifying the unknowns (luck or hidden information) is what probabilistic thinking is broadly about. As nothing is ever completely black or white, good investors know that you cannot have a 100% conviction in your decision. Instead, you can be ‘‘very’’ confident or “fairly” sure, leaving the room for luck and hidden information to play out their own part.

“When we think probabilistically, we are less likely to use adverse results alone as proof that we made a decision error, because we recognize the possibility that the decision might have been good but luck and/or incomplete information (and a sample size of one) intervened. Maybe we made the best decisions from a set of unappealing choices, none of which were likely to turn out well. Maybe we committed our resources on a long shot because the payout more than compensated for the risk, but the long shot didn’t come in this time. Maybe we made the best choice based on the available information, but decisive information was hidden and we could not have known about it. Maybe we chose a path with very high likelihood of success and got unlucky. . .”

There are two identifiable components in probabilistic thinking that are centered on notions of what we consider certain and uncertain.

Certain, uncertain and changes

The things we know can be largely attributed to our beliefs, which are a function of: a) the past, b) our current exposure to new sources of information, and c) how we choose to think about a) and b). The past is not only a repository of what we have learned in the past, but also the foundation of all biases, which become more ingrained as we grow older. Our current exposure to new sources comes from our research, such as reading articles, blogs, having conversations with others, and digging online.

The last component shaping our beliefs relates to thinking and analyzing the data points we have gathered so far and forming our own views and opinions. This can be tricky because it involves blocking out the influence of others in the process of developing our own thoughts.

How deep can we go? For example, if we are analyzing an oil company, we might begin by looking at the company’s financials, their business model, studying the profiles of the management team, reading the latest presentation and annual report, and listening to their latest earnings call. We might then go deeper and study other players in the industry, the supply chains, and the macro dynamics. We might organize a call with the company to answer our questions. We might even delve deeper into understanding the history of the entire industry and the geopolitics of the countries where they do business. However, we may realize that this opens up many more questions. At some point, we need to stop digging and acknowledge that we are leaving some questions unanswered, some unassessed. There is only so much we can research.

Acknowledging what we don’t know also means being comfortable with saying to ourselves and others, “I am not sure.” In modern society, those words tend to have negative connotations. There is pressure to always have an answer.

The other aspect contributing to uncertainty is that things are constantly changing. All the research we conducted on that oil company a year ago will need to be updated with new information. As a result, our own beliefs and opinions should also be updated and challenged (see Paul Saffo’s argument: Strong Opinions, Weakly Held). It is not easy to let go of our most deeply ingrained beliefs, especially those we have held for a long time, those linked to our identity, or those that we have been publicly discussing for years. It is also not possible to be “on top” of all the topics we have opinions on.

Putting it all together

The key ingredient for making good decisions (in the casino, investing, or in life) is to have a process that is immune to outcome. The process involves mapping out the unknown and assigning a probability value to its materialization. This probability value is highly dependent on the quality of the inputs you are exposed to.

What makes a decision great is not that it has a great outcome. A great decision is the result of a good process, and that process must include an attempt to accurately represent our own state of knowledge. That state of knowledge, in turn, is some variation of ‘’I am not sure’.

Over the long term, superior skills, including the assessment of things you do not know and things attributed to good luck, will overcome the impact of bad luck. While a lucky golfer may beat Tiger Woods by shooting a hole in one, they probably won’t beat him over the long term (in this case, 18 holes played four days in row).