Noise: Overcoming the flaw in human judgment

Kahneman and his co-authors’ book is a study of human error in judgments. There are two components of errors in judgment. “Bias and noise – systematic deviation and random scatter – are different components of error.”

Biases are systematic errors of judgments and capture the psychological mechanisms leading to error, such as confidence bias and recency error, these also include the suggestion of favouring groups or interests such as racial bias or political opinion. You could suggest that these are predictable.

Noise on the other hand is not systematic and therefore not predictable. These are variability in decisions that may depend on the day of the week, or the mood of the decision maker and are erratic. Consider two similar judges who give different sentences to people who have committed matching crimes. Or those same judges making different decisions depending on whether it is Friday or Monday, or if they have just eaten lunch. 

You could also consider the same differences in decisions for matching cases by different doctors assessing matching patients and for different outcomes by insurance assessors considering matching claims.

Variability in judgments, that should be identical, is noise. According to the book’s authors, “Wherever you look at human judgments, you are likely to find noise. To improve the quality of our judgments, we need to overcome noise as well as bias.”

Investors are not immune from noise, but early in the book, Kahneman and co-authors we think unwittingly suggest a way of overcoming it.

The authors refer to mechanical prediction, a set of binary rules to achieve an outcome. Mechanical prediction, according to the book’s authors, is superior to people. Sure, people outperform the simple rules now and again, but ‘noise’ prevents them from doing it over and over again. Over the long term, a simple algorithm outperforms human judgment.

This echoes S&P’s half-yearly, SPIVA scorecards. We think the simple rules are market capitalisation indices such as the S&P/ASX 200 that only include constituents based on size. If a company is big enough, it gets in.

Time and time again, human judges, i.e. active managers underperform the ‘mechanical’ benchmark. Over the long term, as more ‘noise’ clouds judgments, the number of managers that outperform the simple rule continues to drop.

“The simplest rules and algorithms have big advantages over human judges: they are free of noise, and they do not attempt to apply complex, usually invalid insights about the predictors.” According to Kahneman and his co-authors.

But the book goes a step further, analysing more complex models and AI.

Applying this to investing, more complex evaluation would involve analysing large data sets, say company’s balance sheets and annual reports, and then making rules to capture those companies with common patterns. 

You could capture those companies with the highest return on equity, lowest leverage and stable earnings growth. These are characteristics of companies that exhibit the ‘quality’ factor, and these more ‘complex’ models are the foundation of smart beta investing.

Rules can be created for other identifiable investment ‘factors’ such as value, low size and momentum. And there are smart beta ETFs on ASX that utilise these more complex algorithms.

Again, ‘noise’ is eliminated from the decisions. The advantage of more complex rules, according to Kahneman and his co-authors “is not just the absence of noise but also the ability to exploit much more information.”

The authors of Noise ponder, “Given these advantages and the massive amount of evidence supporting them, it is worth asking why algorithms are not used much more extensively.”

We agree, we think smart beta has the potential to help investors improve portfolio outcomes. The question the authors of Noise put to readers is, “the algorithm makes mistakes, of course. But if human judges make even more mistakes, whom should we trust?”

We think, smart beta ETFs can help investors eliminate ‘noise’ and help them achieve their investment goals.