My THoughts about uncertainty
By: Robert Wagner
February 21, 2017
In this short introductory note, I would like to share some of my observations about uncertainty and it's linkages to risk management. While the thought process and risk management framework I came to rely on includes numerous steps and components, underneath it is the basic appreciation that almost every decision involves uncertainty. No matter how convincingly or neatly the supporting analysis or argument is being presented, certainty usually is at best an illusion. Whether evaluating trading strategies, investment alternatives, building a team, or allocating capital, uncertainty involved with these decisions needs to be understood and managed.
In order to think about uncertainty, it helps me to first contemplate the meaning of certainty, its exact opposite. What do we know for certain? Are there things we can determine with one hundred percent accuracy, or does nearly everything involve some degree of variability and risk?
First, we can account for some very simple situations with certainty. I can say that I am one hundred percent certain that as I look out the window, my two dogs Oscar and Chips are playing in the yard. I am also one hundred percent sure it has snowed yesterday. Moreover, I am one hundred percent sure that two plus two equals four.
From these simple examples, some common traits emerge, which provide rough guidelines for how an individual may experience certainty. Things, simple facts that we can observe with our own eyes can perhaps be pronounced certain (or almost - maybe the dogs were fighting, not playing?). This may extend to facts or events that occurred in the past, although our memory tends to fade fast and past events often go through mental revisions with the passing of time. Even the traditional concept of facts seems to have developed in strange ways in recent times, seemingly reducing the certainty associated with them.
Taking some of the things we may know for certain, and adding a few layers of complexity begins to muddy the waters. Any statement involving the future by definition involves a degree of uncertainty. Will my dogs play again tomorrow (they have played every day of their lives so far)? Will it snow again tomorrow? Will capital markets and the world order one hundred years from now resemble that of today? These questions have no certain answers, only opinions and some potential estimates of likely outcomes and probabilities.
I would argue that uncertainly applies not only to the future but also to our understanding of the present and the past. Thanks to chroniclers of history and archeologists we have a decent understanding of what ancient civilizations may have looked like. We tend to know more about Greeks and Romans than the later Dark Ages, but even this knowledge involves many theories, inferences, and opinions. The general themes seem more certain and based on fact, but as we get more specific to what happened to any one individual or event, the chances are we have a mix of fact and theories involved.
Even when it comes to interpreting recent or current events, opinions tend to be varied in terms of explaining cause and effect, what really happened and why. My guess is that much of these explanations may be wrong, some just a little, some completely, although the people providing these opinions seem to believe they know the answers with certainty with little room for "what-ifs" and fallibility.
Uncertainty also applies to observations of larger phenomena which we obtain through sampling procedures. Polls are a prime example, where a small sample is selected in the hope of representing characteristics of a population. A large amount of statistical knowledge and techniques relates to sampling, probabilistic evaluation of accuracy, and trying to obtain inferences of populations from limited samples. None of these are certain, there is plenty of room for uncertainty.
Certainty is also hard to come by when it comes to seemingly exact processes such as accounting. A typical corporate financial statement includes so many estimates and judgment that the resulting figures may appear exact, but may be anything but certain. As many figures involve estimates of future outcomes, or measurements through samples, the numbers are built from a mix of fact and expectations.
There are areas where certainty, uncertainty and risk come together. The simple 6 sided dice has a near certainty of landing on one of its six sides (there is a tiny chance it may land on its edge or fall into a wine glass). The chance of its landing on any one side is of course 1/6 and if the dice is fair, these chances will come to realization after many rolls. The uncertainty of where the dice will land on any particular roll creates risk for those betting on a particular outcome. Similar to rolling a dice, there are many games and processes where the number of possible combinations of outcomes are limited, and probabilities can therefore be evaluated with a good degree of accuracy. In these instances, while there is uncertainty of outcome on any one trial, risk can be quantified with a good degree of accuracy.
Most aspects of life involve some degree of uncertainty. It took me months of examples and some interesting conversations to explain this idea to my two kids, aged 8 and 10. They would argue with me endlessly that some things were really one hundred percent. We know the Sun will come up tomorrow is one hundred percent sure, right? On the way to a movie, we know one hundred percent we will get there even if we are a bit late... well, lets hope so... and many, many other examples. We eventually came to agree that few things are ever one hundred percent. However far-fetched a scenario may seem, there may be odds that start to make sense especially over a longer time-horizon. Nowadays, when I tell them something is "for sure", they almost immediately come back at me with a "but it is not one hundred percent"! I think they are on the right track...
An appreciation of the ways uncertainty needs to be figured into many decisions can be a real advantage versus holding strong convictions about things that can turn out to be false. In my experience, a large segment of people do not distinguish between zero probability and a small probability nor at the other extreme between something that is viewed as one hundred percent certain vs. maybe just a small chance of it not being so. The probability of a hundred-year flood on any given day is small... in a given year it would be about one percent (still small), but over several decades the chances will be much higher. The tiny initial difference between these views can lead to huge difference in the evaluation of risks, planning for those risks, and the eventual outcomes. Taking few things for granted and constantly questioning assumptions is not a bad mindset when it comes to serious thinking about risks.