Tuesday, May 3, 2011

Learning From Random Events

Humans have the ability to learn from their mistakes. In so doing, they can reduce the likelihood of repeating such mistakes, which should theoretically lead to a better existence. Therefore, it should come as no surprise that investors wishing to improve their investing acumen will attempt post-mortems on their investment decisions in order to determine what went right and what went wrong. But as I attempted such a post-mortem last week on a risky investment that didn't pay off, a commenter made an interesting point:

"I think one has to be *very* careful about learning lessons from random outcomes...investing is an inherently probabilistic activity...there's a huge impact from the unpredictable and unknowable."

The commenter then went on to point out another risky investment that this time did pay off, and made the point that the outcomes of these two cases could have easily been the reverse (i.e. the first stock could have paid off while the second stock failed), based on the information that was public at the time.

I think the commenter makes a valid point to a certain extent. The problem with such post-mortems is that even though things turned out a certain way, we cannot know whether this was the result of an unobserved or unobservable phenomenon which skewed the results a certain way. Maybe the reasons for the investment were poor, for example, but the investment still worked out because market prices rose as a result of unforeseen economic events. In this case, the investor would learn a false lesson that may not help or may even hurt his attempt to become a better investor.

However, this does not mean studying past investments is useless! If done on a large scale, I would argue that it's the best way to learn how to invest, as it utilizes one of the best forms of learning there is: the case method.

Statistical research demonstrating the effect of one or two factors (e.g. the effect of low P/E and/or management ownership on future returns) at a time can be useful, but only by studying many cases where an infinite number of these factors interact with each other can the investor develop his skills at the art of investing.

By studying only a handful of cases, the investor could be led down the wrong path due to the effects of randomness. However, the investor need not limit himself to only the handful of cases he has himself experienced. The works of numerous value investors are published and available to all. Investors who take advantage of the many thousands of cases that are out there gain the experiences of seasoned investor veterans.


Anonymous said...


Yes, I agree with what you said 100%. If an investor learns not only from how his own investments turned out, but instead he studies a large number of similar cases to see how such cases turn out *on average*, then the lessons learned will be much more robust. By averaging a large number of independent random outcomes together, the "signal" can be extracted from the "noise".

Unfortunately that's much easier said than done. So I think an important element here is *self awareness*. Be aware that due to its highly probabilistic nature, investing is unlike almost all other human activities, so we need to be skeptical of our gut instinct to learn too much from a small number of negative or positive personal investment experiences. In this sense it's similar to other areas of Behavioral Finance, where common human behaviors wind up hurting long-term investment performance.

In a recent letter to investors, Whitney Tilson discusses this issue of randomness. He describes his logic for investing in LECG, but that unfortunately the stock declined 80% in a single day. He writes,

"In light of this permanent loss of capital, why aren’t we certain that this was a mistake, as Netflix clearly was? Because it’s possible that we made a high-expected-value bet, but just got unlucky. Investing is a probabilistic business so it does not necessarily follow that every time you lose money, you made a mistake (and, conversely, every time you make money, you made a good investment). This is very simple and, to us, obvious, but is very poorly understood."

- aagold

Assaf Nathan said...

Hi Saj,

I am reading your blog for a long time now.

For me, this is one of your best posts and one of the most important.

Maybe you should put it in the homepage :)

Saj Karsan said...

Thanks, Assaf!

Anonymous said...

As a poker player this post resonates with me very well. In poker most of the players who simply learn from outcomes tend to evolve into bad players. In contrast, the players who analyze their decisions in terms of expected value can identify mistakes regardless of their outcome, and eventually become good players. The same hazards of result-oriented thinking definitely exist in investing.

It's admirable that you're NOT ONLY self-critical enough to write publicly about an investing mistake you made, BUT ALSO open minded to the point where you learn from the analysis of the mistake. Most people just take a stance and stick to it. Self critical analysis is the right way to look at investment decisions.