Explanations over predictions

The human mind is wired to seek patterns, draw conclusions, and make predictions based on observed data. Whether it’s predicting the weather, investing in the stock market, or understanding natural phenomena, we rely on induction—the process of generalizing from specific instances. But lurking beneath this seemingly straightforward approach lies a profound philosophical challenge: the Problem of Induction. The Problem of Induction, famously articulated by philosopher David Hume, questions the validity of drawing universal conclusions from limited observations. It challenges our intuitive belief that the future will resemble the past. Let’s explore this paradox.

The Inductive Leap

Imagine a turkey on a farm. Every morning, the farmer feeds the turkey. The turkey observes this routine for months. From its perspective, the inductive reasoning goes like this:

“Every morning, I get fed.”

“Therefore, I will get fed tomorrow.”

But then comes Thanksgiving. The turkey’s inductive prediction fails catastrophically. The farmer’s intentions were not as benevolent as the turkey assumed. This simple example illustrates the heart of the problem: Inductive reasoning assumes continuity, but reality can surprise us.

The Fallacy of Extrapolation

Induction often involves extrapolation—extending observed patterns into the future. We assume that the sun will rise tomorrow because it has risen every day in our memory. But what if the sun suddenly decides to take a day off? Our inductive leap becomes a fallacy.

Bernard Madoff, the mastermind behind one of history’s largest Ponzi schemes, managed to deceive investors for years. Despite warnings from financial experts like Harry Markopolos and Ed Thorp, Madoff’s alleged $65 billion fraud persisted. Markopolos, a self-confessed math geek, meticulously documented red flags and sent detailed memos to the Securities and Exchange Commission (SEC), urging them to investigate. However, Madoff’s consistent monthly returns—whether achieved through dubious means or the Ponzi scheme—lulled investors into complacency.

Harry Markopolos found it impossible to replicate Madoff’s returns using standard financial models. The complexity of Madoff’s strategy made it difficult to verify its legitimacy. Madoff claimed to use a split-strike conversion strategy involving S&P 100 index options. Markopolos found this explanation implausible and suspected it was a cover for the Ponzi scheme. He couldn’t find a plausible, sound and replicable strategy. Hence, he became suspicious and asked questions. “Sanity calls for suspicion in face of extraordinary claims, stupidity calls for awe.”

The allure of steady profits blinded investors to the impending collapse of Bernie Madoff’s Ponzi scheme. They couldn’t argue against the allure of ‘what-if’ Bernie’s returns were actually true?

It reminds me of Chatur from 3 idiots saying: “Matlab nahi chahiye Dubeyji, mein yadd karlenge. I will memorize it”

The Black Swan & Falsifiability

Nassim Nicholas Taleb introduced the concept of the “Black Swan.” A black swan is an unexpected, rare event that disrupts our inductive expectations. Before the discovery of black swans in Australia, Europeans believed all swans were white. The unexpected sighting shattered their inductive assumption. Science grapples with the Problem of Induction. Scientists collect data, formulate hypotheses, and test them. But how can we justify the leap from observed data to universal laws? Karl Popper, another influential philosopher, proposed a solution: falsifiability. Scientific theories should be testable and potentially falsifiable. A single black swan can disprove a theory. Think about falsifiability in this way – Whereas an incorrect prediction automatically renders the underlying explanation unsatisfactory, a correct prediction says nothing at all about the underlying explanation.

Just like a scientific theory can be falsified by new evidence, the assumption of high historical returns for a country’s equity index can be falsified by poor performance over a long period.

In this case, the theory is that past performance guarantees future results, specifically high real returns. The data (observational evidence) showing the equity index failing to deliver double-digit real returns over 30 years acts as falsification. This falsification suggests investors should adjust their expectations and consider the possibility of lower real returns, even mid-single digit returns being rare. This aligns with the idea that falsifiable approaches in science lead to revising expectations in light of new information.

Here are two of many extrapolations that I have heard recently:

1. Flows and liquidity wouldn’t allow stocks to fall.
   • Falsifiability: At most market peaks, flows, liquidity and sentiment is rock solid. Stocks can fall just because incremental buyers become noncommittal. Past flows can’t protect future crashes.
     
     
2. Economic reforms and growth equals equity returns.
   • Falsifiability: The period of transformational reforms in India, from 1991 to 2001, Sensex Index produced 8.6% nominal returns, barely beating inflation or bonds.

You would encounter many more such extrapolations.

Richard Feynman once said:

“The first principle is that you must not fool yourself—and you are the easiest person to fool.”

Feynman reminds us that our cognitive biases, including inductive leaps, can lead us astray. Scientific method of processes and explanations, that root-cause identification, attempts to minimize this self-deception.

Here’s what to do:

Challenge the assumption of high returns: Don’t take for granted that your investments inherently deserve high returns. Ask yourself why they should outperform the market.

Focus on falsifiable expectations: Instead of predicting high returns based on past performance, consider what factors could falsify that expectation.

Limited use of past performance: Past performance, even if consistent for a few years, is not a scientific guarantee of future results. Markets are complex and can change significantly.

Seek explanation not extrapolations.