Instinct To Analytics: A Decision Making Spectrum

I’ve read several articles debating the merits and deficits of the two different schools of decision theory. The debate tends to be either/or but I believe it should be both. Let me explain with a few stories:

INSTINCT->RIGHT. A fire chief walks into a routine kitchen fire and immediately pauses. He tells his team to hold and then yells for them to get out as fast as they can. Just a few seconds after they exit, the floor under the kitchen collapses. (paraphrased from “Seeing What Others Don’t” by Gary Klein)

ANALYTICAL MODEL->RIGHT. A group of doctors was asked to analyze the biopsies of 193 Hodgkin’s disease patients and predict the survival time of each patient. Their correlation with actual survival times were effectively 0, meaning the doctors' forecasts had no predictive power. However, constructing a simple linear model using the variables the doctors themselves labeled as important on the biopsy, you can accurately predict survival time. (paraphrased from “The Robust Beauty of Improper Linear Models in Decision Making” by Robyn Dawes)

INSTINCT->WRONG. A young musician gets his demo tape in front of a senior record company executive who loves the young musician’s sound. The tape makes it around the expert syndicate and receives overwhelming praise, ultimately reaching U2’s Executive Producer who said this musician was going to change music. The problem? The public didn’t agree with the experts and the musician’s career never came close to the heights predicted by the experts. (paraphrased from “Blink” by Malcolm Gladwell)

ANALYTICAL MODEL->WRONG. A group of mathematicians sought to build a single variable measure of risk for financial instruments. The measure, Value at Risk, became widely adopted across the financial community and on its back, trillions of dollars of poorly priced products were bought and sold, leading to the financial crisis of ’08.

As you can see from these stories (and there are thousands more), there is no one right way to make decisions. It depends on the situation. In Decision Theory, two schools of thought have emerged. One that trusts instinct (leaders include Gary Klein, Gerg Gigerenzer and was made popular by Malcolm Gladwell’s “Blink”) and another that does not trust instinct (leaders include Daniel Khaneman, Amos Tversky, Dan Ariely, Ron Howard and has been made popular in investing as topics called Behavioral Finance and Neuroeconomics). The basic distinction could be boiled down into each group’s belief in mankind’s rationality. I suspect, however, that the experts in the field would agree that the ultimate solution depends on the decision making situation and that the lines between their philosophies are not so black and white.

Even the nomenclature is blurred. I myself have lumped gut instinct and heuristics together, but they are actually quite different (see the spectrum below). For instance, if I were to ask a portfolio manager why he invested in a stock and he said, “I like the story”, that could be considered instinct. If the portfolio manager said, “I like the story and I only invest in companies that trade below 15x earnings and where I’ve met the management team” then the PM would be using a combination of Instinct and Heuristics (rules of thumb).

In fact, I would suggest that there is a continuum from pure instinct to purged of all instinct (Analytics): Instinct and Heuristics (generally associated to the Instinct camp) to Logic and Analytics (generally associated with the anti-instinct camp).

To continue the portfolio manager analogy, if the decision process has explicit inputs like price targets, probabilities, conviction levels that result in a specific buy/sell decision then logic is being applied. In a purely analytical environment, any element that involved subjectivity is removed like (i.e. conviction level or subjective probabilities).

As anything in life, the extremes are rarely correct. Pure instinct is dangerous and ignores troves of psychological analysis that shows how poorly designed humans are at making complex financial decisions. Pure analytics assumes that humans are idiots that have instincts that are wrong or at least random. I’d suggest the truth is somewhere in between.

A good decision process removes instinct from the decision driver’s seat but keeps it as a meaningful input that influences the ultimate outcome of the logical process. Conviction level and subjective probabilities are good examples. Subjective ratings of management teams, balance sheets, etc. are also logical inputs. Good process can involve smart heuristics like never investing in companies that have market caps greater than their market size or shorting companies with Short Interest Ratios above 40%. It is important to capture these heuristics in checklist form so that they are properly accounted for in the output. What you end up with is an Analytic model that draws from the Instinct of the experts. This becomes the engine of the decision process and can be refined through observation, instinct, and data analysis (feedback from the model).