Investors Should Embrace Probabilities to Improve Decision Making

One of the many oddities of the investment industry is that whilst it operates in a constant state of uncertainty and flux, its participants talk in the language of certainty.  Confident forecasts and dogmatic opinions are valued far more highly than circumspection and caveats.  This bias stems from the mistaken belief that expertise is related to conviction – that an expert must know the ‘right answer’.  Whilst this might hold in certain situations where the environment is stable and skill dominates outcomes; the reverse is true in random and unpredictable domains, such as financial markets. Here, expertise is more evident in humility, a willingness to revise views and to deal in probabilities.

There is certainly truth to the view that we are poor when it comes to thinking in terms of probabilities, and there are multiple examples of our irrationalities in this area – such as the tendency to neglect probabilities when considering extreme scenarios, and our propensity to overweight or ignore small probabilities.  The notion, however, that we should not talk about probability because we have limitations in this regard is entirely spurious; we cannot make a decision without taking some view on the likelihood of potential outcomes, even if we are not explicit about it.  If we bring our probability assessments into the open; it materially improves our ability to understand our biases, receive feedback on our judgements and learn.

Aside from our general struggle with thinking in such a manner, there are a range of other issues that limit the use of probabilities when making investment decisions and forecasts:

 Lack of expertise: As soon as probabilities are expressed around a variety of potential outcomes, there is an acknowledgement of uncertainty, which is often erroneously viewed as a lack of expertise and akin to the cardinal sin of stating “I don’t know”.  This issue is exacerbated by the fact that there will be others making bold, singular predictions – surely, they must know better?

Spurious accuracy: In uncertain environments views on probabilities are necessarily based on subjective judgements. Applying specific probabilities to a range of scenarios can appear overly scientific – what does it mean to believe the likelihood of an event is 17%?  This view, however, misses the value of applying probabilities.  Its use is not as a precise figure but as a measure of confidence and a means to monitor how our views evolve through time.

Implementation challenges: It is far easier to translate a strident, narrow view into an investment decision, than to attempt to reflect uncertainty and a variety of potential outcomes.

The forty percent problem:  Probabilities can be misused when making forecasts – the classic case is an individual assigning a 40% probability to an outlier event (a recession, typically). The 40% level means that if the event doesn’t occur their forecast was correct (on balance), but if the scenario does transpire it is of a high enough likelihood for them to be feted for the prediction.  This is an example of probabilities being utilised in a strategic, unhelpful fashion.

Changing minds:  When we frame outcomes in terms of probabilities, we allow ourselves the freedom to alter our view.  Whilst this should be considered positive and a reflection of realistic pragmatism; it is more often stigmatised as ‘sitting on the fence’ or providing a pre-prepared excuse for ‘being wrong’.

None of these arguments are particularly compelling and they are outweighed by the manifold benefits of thinking explicitly in terms of probabilities when making investment decisions:

Humility:  Simply by framing potential outcomes in terms of probabilities, we immediately acknowledge the uncertainty and unpredictability of any given judgement. We are divorcing ourselves from hubristic, narrow forecasts from the outset.

Reduce confirmation and commitment bias: When we espouse high conviction and overly precise views we come to be defined by them. Instead of refining or rejecting our beliefs in light of contrary evidence, we instead seek to preserve our ego by becoming increasingly committed and ignoring countervailing information.  By ascribing probabilities to a range of different scenarios we are more likely to remain open-minded and diminish the influence of being committed to any particular outcome.

Understand strength of view: It is often difficult to gauge the strength of someone’s view – this is particularly the case for binary situations where a ‘yes’ decision could either represent a marginal call or be a display of resounding confidence.  For example, if we had to predict the result of two coin tosses one, which was biased 55% in favour or tails and the other 95% towards tails – our ultimate view on the most likely outcome should be the same – but the level of confidence wildly different. Given that the investment industry tends to favour those making binary, conviction calls; such nuances are often lost and marked uncertainty can exit beneath a veneer of confidence – employing probabilities can help to reveal this.

Consider alternative scenarios: The use of probabilities forces us to consider alternative (often negative) scenarios; an individual with a high level of confidence in a specific outcome is still likely to ascribe some probability to other less favourable results.  Even if the chance of these occurring is regarded as minimal, there is an undoubted benefit from considering them, rather than ignoring them entirely and focusing solely on the primary case.  The best example of this would be in assigning a probability of a bullish case for a particular stock – the simple act of not giving the positive scenario 100% likelihood, compels consideration of other potential outcomes, and hopefully encourages debate.

Incorporate new information: A crucial element of successful decision making is the ability and willingness to revise views in the light of new information. If we make binary decisions (such as buy or sell) we can easily avoid incorporating fresh evidence by claiming our view has not altered (“it is still a buy”); however, by specifying probabilities it becomes increasingly difficult to ignore developments that are likely to impact your level of confidence.

For example, imagine you buy an active mutual fund and hold a 70% confidence that it will outperform its benchmark over three years; after one year the fund has generated significant excess returns – do you revise your view? If it is a simple, vague buy decision you are likely to continue to hold; however, if you have to update your confidence level there is likely to be an impact. Given the evidence of mean reversion in active manager returns, can you really still can you really continue to ascribe the same probability of outperformance over the next three years, or should it be revised lower?  Being explicit with probabilities forces us, and others, to challenge and update our thinking.

Evaluate past decisions: Given the scourge of hindsight bias it is often impossible to assess the quality of our historic decisions, as we cannot accurately recall our thought processes and feelings at the particular time. If, however, we consistently review and document our decision / forecast confidence level, we are in a far better position to understand how we came to a particular viewpoint and the manner in which we reacted to new information.  This can become a vital tool in learning from past behaviours.

Ascribing likelihoods to potential outcomes is not easy, as with all our behaviours it will be blighted by noise and bias; however, even if we are not openly expressing opinion in probabilistic terms, they are still deeply embedded in the views we hold.  As highlighted in Philip Tetlock’s Superforecasters and Annie Duke’s Thinking in Bets, being explicit about probabilities when making judgements or forecasts allows us to embrace uncertainty, affords us the freedom to revise our opinion as new information arrives and fosters the ability to learn from previous decisions.  In complex and ambiguous financial markets such features are invaluable.