Why Humans Need Explanations for Randomness
Pattern Recognition, Cognitive Biases, and the Search for Meaning in Uncertain Environments
Human beings are remarkably uncomfortable with randomness. When an event occurs, we instinctively search for a reason. When a market rises, investors want to know why; when a company suddenly collapses, analysts construct narratives explaining what went wrong; and when an unexpected outcome appears, we retrospectively assemble the sequence of events that supposedly made it inevitable.
This impulse is deeply embedded in human cognition. The mind is designed to detect patterns, infer causes, construct narratives, and reduce uncertainty. These abilities are extraordinarily useful because they allow humans to learn from experience, anticipate threats, coordinate with one another, and make decisions in environments where information is incomplete. Yet the same mechanisms that help us navigate uncertainty can also create systematic errors when the underlying environment contains genuine randomness.
Randomness is difficult to accept because it undermines the assumption that events should have coherent explanations. A genuinely stochastic process can produce outcomes that look meaningful without being generated by a meaningful underlying pattern. A sequence can appear structured even when its components are independent, while an extreme event can seem retrospectively obvious despite having been extremely difficult to anticipate beforehand.
Financial markets provide an especially fertile environment for this problem. Prices move continuously, information arrives unevenly, participants interpret events differently, and the consequences of decisions are observable. This creates an almost irresistible demand for explanation. The result is one of the central paradoxes of investment analysis:
the more information becomes available, the easier it can become to construct convincing explanations for events that may have contained a substantial element of randomness
Understanding why humans need explanations for randomness is therefore, not merely a question of psychology. It is also a question of how investors interpret evidence, construct beliefs, assign causality, and make decisions under uncertainty.
The Human Search for Causality
Human cognition is fundamentally causal. We do not merely observe that one event followed another; we instinctively ask whether the first event caused the second. This tendency is adaptive because recognising genuine relationships can protect us from danger and allow us to make better decisions. If a particular plant repeatedly makes someone ill, learning the relationship can prevent future harm. If a particular sound consistently precedes danger, recognising the association can improve the likelihood of responding appropriately.
The problem arises when the environment contains relationships that are weaker, more complicated, or entirely coincidental. Humans can perceive causality where none exists because the cognitive system is highly sensitive to patterns and associations. In evolutionary terms, the occasional false positive may have been less costly than failing to recognise a genuine threat.
This produces a powerful tendency toward pattern detection. A sequence of random events can therefore appear structured, while a cluster of unusual outcomes can feel too meaningful to be accidental. The mind tends to prefer an explanation to an absence of explanation, even when the available evidence does not justify one.
The Discomfort of Not Knowing
Randomness creates a particularly difficult psychological state because it imposes uncertainty without necessarily offering an identifiable mechanism. If a machine fails because a component has broken, there is a concrete explanation that can potentially be investigated. If a company misses earnings because a specific contract was cancelled, the event can be incorporated into a causal narrative that links the outcome to an observable event.
The situation becomes considerably more uncomfortable when a market moves sharply without any clearly identifiable new information. Investors may begin searching for explanations, wondering whether institutional positioning changed, whether an algorithm triggered selling, whether geopolitical concerns emerged, or whether some participants possessed information that others did not. Some of these explanations may be correct, but others may simply be stories constructed after the fact to make an uncertain event feel more intelligible.
This distinction is critical because an explanation can be psychologically satisfying without being empirically valid. The human desire to resolve uncertainty can therefore encourage premature conclusions precisely when uncertainty should remain explicit.
Apophenia and Pattern Recognition
The tendency to perceive meaningful connections in unrelated information is often described as apophenia. Pattern recognition itself is not a flaw; indeed, it is one of the most important capabilities of human intelligence. The problem occurs when pattern recognition exceeds the amount of genuine structure contained within the available data.
Financial markets are particularly susceptible because they generate enormous quantities of observations. Prices, volumes, earnings, macroeconomic statistics, sentiment indicators, analyst revisions, news events, options activity, and alternative datasets can all be examined simultaneously. With enough variables and enough possible relationships, apparent patterns are almost guaranteed to emerge.
A relationship discovered after searching thousands of possible combinations may therefore, look highly persuasive even if it occurred by chance. This is one reason statistical significance must be considered alongside economic reasoning; as a relationship that appears in historical data does not necessarily represent a durable mechanism, particularly if there is no convincing explanation for why the relationship should persist.
The challenge is therefore not simply finding patterns; instead, it is determining which patterns contain information about the underlying system and which are artefacts of randomness, selection, or excessive searching.
The Narrative Fallacy
Humans prefer stories to distributions because stories organise complexity into a form that is easy to understand and remember. A narrative has characters, causes, consequences, and a beginning and an end, whereas a probability distribution simply describes a range of possible outcomes and their relative likelihoods.
After an event occurs, this narrative instinct becomes especially powerful. Suppose a stock falls by 30 percent. An analyst can usually identify a sequence of developments that appears to explain the decline, including slowing growth, weaker consumer confidence, rising interest rates, an earnings revision, or a management decision. Such an explanation may be entirely plausible, but plausibility alone does not establish that it was the primary cause of the movement or that it could have reliably predicted the outcome beforehand.
This distinction lies at the heart of hindsight bias; the fact that a coherent narrative can be constructed after an event does not mean that the same narrative possessed predictive power before the event occurred. The historical record contains the outcome, while the decision-maker operating in real time had to confront a distribution of possible outcomes.
Investment analysis becomes less reliable when those two perspectives are treated as equivalent.
Hindsight and the Illusion of Predictability
Once an event has occurred, its uncertainty disappears from the historical record. The event is now known, and this knowledge changes how the preceding circumstances are interpreted. Developments that previously appeared ambiguous can suddenly seem like obvious warning signs, while alternative possibilities that were plausible at the time disappear from the narrative.
Financial commentary frequently illustrates this phenomenon. After a major market movement, explanations appear rapidly as analysts identify catalysts, commentators interpret investor psychology, and historical precedents are brought forward. The resulting narrative can make the event appear almost inevitable even though, before it occurred, several different outcomes may have been entirely plausible.
This is why investment research should distinguish between explaining an event and predicting it. A compelling explanation of yesterday does not automatically constitute a useful model for tomorrow.
The distinction is particularly important when evaluating investment skill. An investor who makes a correct forecast for the wrong reasons may appear more capable than an investor whose carefully constructed probabilistic forecast happened to produce an unfavourable outcome. Observed outcomes therefore provide incomplete information about the quality of the underlying decision process.
The Representativeness Heuristic
One reason humans struggle with randomness is the representativeness heuristic, through which people often expect small samples to resemble the broader characteristics of the population from which they are drawn.
Consider a fair coin. A sequence of ten heads may feel less random than a sequence that alternates neatly between heads and tails, even though both sequences are perfectly compatible with the same random process. The alternating sequence may appear more representative of randomness because it looks balanced, while the sequence of repeated heads feels as though something unusual must have happened.
This intuition is misleading because randomness does not necessarily produce outcomes that look random. Clusters, streaks, repetitions, and extremes can all emerge naturally from random processes.
The same principle matters in financial markets. An asset can outperform for several consecutive years, a strategy can experience a long period of success, or an individual investor can appear consistently successful. The existence of a streak does not, by itself, establish that the underlying process is non-random or that persistent skill is responsible. The investor must instead determine whether there is a credible mechanism that explains why the observed outcome should continue.
The Gambler's Fallacy
The opposite error is the gambler's fallacy, in which people assume that previous random outcomes alter the probability of future independent outcomes. If a fair coin has produced heads five times in succession, many people intuitively feel that tails is now "due," despite the probability of the next toss remaining unchanged.
Financial markets are obviously not independent coin tosses. Prices exhibit momentum, mean reversion, volatility clustering, liquidity effects, and other forms of dependence. Nevertheless, the underlying psychological error remains relevant because investors can mistake an unusual historical sequence for evidence that a reversal must follow.
The appropriate question is therefore not whether an outcome feels extreme but whether there is a mechanism that changes the probability distribution of future outcomes. An asset that has risen substantially may eventually decline, but the fact that it has already risen does not establish that a reversal is imminent. Likewise, a severely underperforming asset may continue to fall if the forces driving the decline remain intact.
The distinction between psychological expectation and structural probability is essential.
The Clustering Illusion
Random events naturally cluster. If a large number of random outcomes are generated, periods of unusually high activity and periods of unusually low activity will emerge without requiring a structural change in the underlying process.
Humans, however, often interpret clusters as evidence of causality. Three negative events occurring close together can feel qualitatively different from the same three events distributed over a longer period because temporal concentration creates an impression of common cause.
Financial markets frequently amplify this tendency. Several negative headlines can arrive within a short period, multiple companies can miss expectations simultaneously, or several financial institutions can experience stress at roughly the same time. The clustering may reflect a genuine common factor, but it may also arise from correlated responses to a shared environment or from the statistical characteristics of the process itself.
The analytical task is therefore to determine whether a cluster represents information about a changing underlying system or simply an expected feature of uncertainty.
The Law of Small Numbers
Humans frequently draw broad conclusions from limited observations. An investor may experience five successful investments and conclude that a particular strategy works, while a fund that outperforms for three years may quickly acquire a reputation for possessing a superior investment process. Similarly, a company that experiences several quarters of exceptional growth may be assumed to have established a permanently higher growth trajectory.
The problem is that small samples contain substantial noise. An observed result can be generated by skill, luck, or some combination of the two, and distinguishing between these possibilities requires sufficient evidence to estimate the underlying process.
This is particularly difficult in investing because the relevant samples are often inherently small. An investor may make only a limited number of genuinely consequential decisions over a career, while major market regimes occur relatively infrequently. Extreme crises are even rarer, despite having disproportionate effects on long-term outcomes.
Confidence should therefore reflect not only the observed result but also the quantity and quality of evidence supporting the inferred mechanism.
Luck, Skill, and Attribution
One of the most difficult problems in investment analysis is separating skill from luck. Suppose two investors achieve identical returns over five years. It is tempting to conclude that both possess similar investment ability, yet their underlying processes may be entirely different.
One investor may have followed a disciplined strategy with a positive expected value but experienced favourable randomness, while the other may have followed a poor process that happened to produce favourable outcomes. Observed performance is therefore a noisy signal of underlying skill.
The problem becomes even more severe when performance is evaluated over short periods or when a strategy contains substantial exposure to rare events. A process can produce favourable outcomes for years before encountering a particular state of the world that exposes its hidden fragility.
A sophisticated approach to investment evaluation therefore asks not only what happened but how the outcome was generated.
what assumptions were made?
what risks were taken?
what was the distribution of possible outcomes?
was the result consistent with the stated process?
would the decision have remained reasonable if the eventual outcome had been different?
These questions move analysis away from outcome-based storytelling and toward process-based evaluation.
Why Markets Invite Stories
Financial markets create an unusual psychological environment because outcomes are continuously observable. Prices move every second, news arrives constantly, and financial media is under continual pressure to explain what has happened. Investors consequently operate within a permanent demand for narratives.
Every significant market movement appears to require an explanation, yet markets do not necessarily produce a single causal story. Multiple forces can operate simultaneously, some information may be important while other information is irrelevant, and different participants can respond to the same information in opposing ways.
A market price is therefore the result of an interaction between expectations, constraints, information, liquidity, positioning, and individual decisions. Reducing this complexity to one explanation can create false certainty. Sometimes the most accurate answer to the question "Why did the market move?" is that several forces interacted, expectations changed at the margin, liquidity conditions amplified the movement, and a meaningful proportion of the change remains difficult to attribute precisely.
Accepting this possibility is psychologically difficult because it removes the comfort of a neat narrative.
The Seduction of Ex Post Explanation
There is a particular danger in analysing randomness after the fact because once an event has occurred, an almost unlimited number of explanations can be constructed around it.
A market crash can be attributed to valuations, monetary policy, leverage, geopolitics, investor psychology, technology, regulation, or some combination of these factors. Because many of these variables were already present before the crash, it is often possible to construct a convincing retrospective narrative around them.
The relevant question, however, is not simply whether a factor can be associated with the outcome. It is whether the proposed mechanism possessed sufficient explanatory and predictive power before the outcome occurred.
This distinction separates genuine analysis from retrospective storytelling. A useful hypothesis should generate expectations that can potentially be tested. However, if an explanation can accommodate almost any outcome, it may explain everything while predicting nothing.
The Role of Bayesian Thinking
Bayesian reasoning provides a useful framework for managing the human desire for explanation. Rather than treating every new observation as proof of a particular hypothesis, Bayesian thinking asks how much the evidence should change existing beliefs.
An unexpected event does not necessarily invalidate an existing model, just as a successful prediction does not necessarily prove that the model is correct. Instead, new evidence updates the probabilities attached to competing explanations. This creates a more disciplined relationship with uncertainty; an investor can believe that a company possesses a strong competitive position while recognising that the evidence supporting this belief is imperfect. Accordingly, new information can increase or decrease confidence without forcing an immediate binary conclusion.
The same principle applies to randomness. A surprising outcome should prompt an investigation into whether the existing model is incomplete, but that investigation should not automatically become a new causal narrative. The strength of the update should reflect the quality and diagnostic value of the evidence.
Randomness and the Limits of Explanation
There is a deeper philosophical point beneath these behavioural tendencies. Not everything that happens has a sufficiently knowable explanation.
Some systems are stochastic. Others may be deterministic but practically unpredictable because their initial conditions cannot be measured with sufficient precision. In still other cases, the interactions within a system can become so complex that isolating individual causes becomes extremely difficult.
These forms of uncertainty are not identical, but they share an important characteristic:
human beings may not be capable of constructing a complete explanation for every observed outcome
This is difficult for an intelligent species accustomed to solving problems. When faced with uncertainty, the mind often assumes that more analysis will eventually reveal the answer. Sometimes additional analysis does uncover a previously hidden mechanism. In other cases, however, the limiting factor is not insufficient analysis but the structure of the problem itself.
Recognising that boundary is an important component of intellectual humility.
From Prediction to Probabilistic Thinking
The alternative to demanding certainty is not to abandon analysis but to think probabilistically. Rather than asking what will happen, investors can consider which outcomes are plausible, how the available evidence changes their relative probabilities, and how the consequences differ across those scenarios.
This shifts the objective from constructing a single deterministic explanation to understanding a distribution of possibilities; such thinking becomes particularly valuable when outcomes are asymmetric. If an investment has limited downside and substantial upside, the investor may not need to know precisely which outcome will occur. Conversely, when the downside is severe or irreversible, uncertainty becomes considerably more important because being wrong carries greater consequences.
Probability is therefore inseparable from risk. The absence of certainty does not mean the absence of structure; it means that whatever structure exists must be represented honestly.
When Explanation Becomes a Source of Risk
The desire for explanation becomes dangerous when confidence in the explanation exceeds the evidence supporting it. An investor who believes they understand exactly why a market moved may become less receptive to contradictory information, while a manager who constructs a compelling narrative around a company may subsequently interpret new evidence through the lens of that narrative.
The explanation becomes a filter through which subsequent information is processed. This can produce confirmation bias, motivated reasoning, and excessive conviction. The irony is that the cognitive mechanism intended to reduce uncertainty can ultimately, increase decision risk because a coherent story may provide psychological closure without providing informational accuracy.
Good investment processes therefore need mechanisms that preserve uncertainty rather than eliminate it prematurely. A thesis can be strong without being certain, and an explanation can be useful without being complete.
The MorMag Perspective
At MorMag, randomness is treated not as an inconvenience to be explained away but as a fundamental feature of decision-making under uncertainty.
Financial markets contain stochastic variation, incomplete information, endogenous feedback, changing regimes, behavioural responses, and structural relationships that evolve through time. Consequently, not every market movement has a single identifiable cause, and not every successful investment decision demonstrates superior foresight. The distinction between signal and noise is therefore central to investment research. A sophisticated process should seek to identify persistent mechanisms without assuming that every observed pattern represents one. It should distinguish causal relationships from correlations, predictive evidence from retrospective narratives, and genuine structural change from ordinary statistical variation.
This is also why probabilistic thinking matters. The objective is not to construct an explanation capable of making every outcome appear inevitable after it occurs. Instead, the objective is to develop a framework capable of incorporating uncertainty before the outcome occurs and of updating beliefs as new evidence arrives. That distinction changes how investment research should be evaluated. A thesis should not be judged solely by whether its forecast was correct. It should also be evaluated according to the quality of its assumptions, the evidence supporting them, the risks identified, the alternatives considered, and the extent to which uncertainty was represented honestly.
This perspective places an important constraint on quantitative analysis as well. A sufficiently large dataset will almost inevitably contain patterns, which means that the discovery of a relationship is only the beginning of an investigation. The more important questions concern whether the relationship is economically meaningful, statistically robust, causally defensible, and likely to survive changes in the underlying system.
Human beings naturally want markets to make sense. The challenge for investors is not to suppress that instinct entirely because explanation is essential to learning, but to recognise when explanation becomes over-explanation. Sometimes an apparent pattern contains information; sometimes it reflects a structural mechanism; sometimes it is produced by interacting forces. At other times, it is simply what randomness looks like when viewed through a human mind searching for meaning.
Knowing the difference is one of the most difficult problems in investment research.
Conclusion
Humans need explanations because explanations make uncertainty psychologically manageable. They transform events into stories, convert ambiguity into causality, and allow the mind to construct a coherent model of the world. These abilities have been enormously valuable throughout human history, but financial markets expose their limitations particularly clearly.
Randomness can produce patterns, luck can resemble skill, clusters can resemble causality, and successful forecasts can arise from poor processes while sound decisions can produce unfavourable outcomes. Once an event has occurred, the range of possible explanations expands dramatically, making hindsight feel more certain than foresight ever was. The solution is not to abandon explanation but to make explanation more disciplined. Investors must distinguish between what they know and what they infer, between evidence and narrative, and between a mechanism that generates testable expectations and a story that merely fits the historical record.
The deepest form of analytical sophistication may therefore involve becoming comfortable with the possibility that some events cannot be explained with precision. Markets do not owe investors a coherent story, and the absence of a satisfying explanation does not necessarily indicate that the analysis has failed. Sometimes several forces have interacted, information has remained incomplete, and randomness has played a meaningful role in the outcome. Accepting that possibility is not an admission of ignorance, it is recognition of the structure of the problem.
The objective is not to explain everything, instead it is to understand enough of the system to make better decisions while remaining conscious of what cannot be known.
