How'd We Miss That?

The excerpt below is from a book, cited underneath, that is a collection of research studies that were not then concerned with the current popularity of Behavioral Economics. The book is the seed research of that now well regarded topic, and Daniel Kahneman is the father of subject. It has also been further researched by Richard Thaler PhD, among others.
The excerpt addresses, not specifically, but by implication, how we missed the market crashes and how we overweight some scenarios versus other, equally probable ones.

In any plan, the cumulative probability of at least one fatal error could be overwhelmingly high even when the individual cause of failure is negligible.
Plans fail because of ‘surprises;’ occasions on which the unexpected ‘uphill’ change occurs.
The simulation heuristic, which is biased in favor of ‘downhill’ changes,
is therefore associated with a risk of large systematic errors.
In evaluating a scenario, alterations on ‘what could have been done differently’ are many times introduced.
These can be classified as either:
Uphill: a change that introduces unlikely occurrences or surprises
Downhill: a change that removes an unlikely occurrence or surprise
Horizontal: one arbitrary value replaces another in the scenario,
neither arbitrary value is more likely, or less likely
-people are much more likely to undo a scenario with downhill changes than uphill changes…
horizontal changes are almost nonexistent.
Think of a cross country skier. It is easier to ski down than up, the psychological distance from peak to valley is shorter than from valley to peak.
Thus, mental simulations invariably have a preference for downhill variations.
Kahneman and Tversky, Judgment Under Uncertainty

I will explore the topic further in future posts.

Moron Beta

To revisit the current discussion, after my digression, Beta tells you what the relationship WAS between a security and the market. It tells you nothing about what the relationship will be.
As suggested earlier, Beta can be any number you want as it is dependent on the time frame chosen. A one year Beta is one number, and a one week Beta is another for the same security.
Phormula Phreaks argue that if you use the same time frame for both the security and the market, the problem is solved. It isn't. And to make matters worse, frequently the number for Beta is tossed out like it's a constant, and with no reference to the time frame used to calculate the Beta. Beta is a variable, that means it changes. Variability is the mortal enemy of predictability.
In sum, knowing what happened, which is all Beta tells you for a given time frame in the past, is NOT the same as knowing why it happened. Historians and statisticians frequently equate the two. They do this by attributing one off events with causality.
For instance, as pointed out in Everything is Obvious (once you know the answer.) by Duncan Watts, if you notice that each time the wind blows, the leaves of a tree move, that's once kind of cause effect. It happens a lot, it has validity.
What historians frequently do is take a one time event, and attribute causality to subsequent events. That is, if a cat meows, and the leaves shake, the meow caused the subsequent movement in the leaves. It ain't so. His example is a skirmish between English and French ships in the 14th century leading to the hundred years war, as if the skirmish 'caused' the war. Nobody can know that, and certainly can't know it at the time of the sea battle. As it has been put, there is no, "Dear Diary, the hundred years war began today."

Quick Digression

Brit Marling, actor and producer, was featured in the Sunday New York Times Magazine today.
From the article:
The summer before her senior year, though, she took an internship at Goldman. If anything, it left her disillusioned. “I started to feel like it was all a bit of a fraud, all these charts and regressions and models.” She turned down the subsequent job offer...
Wow...she 'gets it' during a summer intern job at Goldman. What's taking everyone else so long? I can answer for myself. I worked in the business for twenty years before I realized predicting the future was crap, voodoo, of no value except to the predictors selling snake oil to the gullible (As I did. My only defense is that I didn't think it was snake oil at the time. I thought analysts, beta coefficients and standard deviations could help. I was wrong.)