Lately I've been reading Gut Feelings: The Intelligence of the Unconscious in which Gerd Gigerenzer argues that certain mental shortcuts can actually provide better judgement than logical or rational processes. What is interesting is his idea of the
Take The Best Heuristic. Basically order the information needed for a decision from most to least important and stop as soon as one has an advantage. The example Gigerenzer provides is predicting the outcome of NBA games based on the teams' performance in the previous season and then the halftime score. If one team did markedly better in the previous season they should be expected to win. If both teams are equal on that measure the team that is ahead at halftime should be expected to win. This method matched the performance of a more specific method that always took the halftime score into account and examined the differences in scores, not simply who was ahead.
Certain variables are useful in predicting the outcomes of uncertain events, but as more variables are added they become less useful and hurt the predictive power. This reminded me of Taleb saying that a simple ratio of liquid assets to debt was a better predictor of corporate default than most experts.
...the expert is the closest thing to a fraud, performing no better than a computer using a single metric, their intuition getting in the way and blinding them. (As an example a computer using a single metric, the radio of liquid assets to debt fares better than the majority of credit analysists.) (The Black Swan, second edition, p. 146)
Looking at a single, important metric does yield good results but Taleb's use of "intuition" seems out of place. I think credit analysts intuitively know that a company with a lot of debt and no cash probably doesn't have a good chance. However, they can't write intuitive research that is only one paragraph. They are supposed to research and find every piece of information about that company because conventional belief is that more information is better. All this extra information doesn't really help though. It's of little or no predictive value and distracts from the important information. Although I agree with Taleb that there is an expert problem and their predictions are bad, it seems to be more of a problem of too much information, not of intuition.
One area I am interested in of the Take The Best method is in the market and specifically technical analysis. The majority of technical indicators and studies for price charts rely on a small amount of information. Moving averages are extremely common and are mostly used to smooth out noise from the price movement. All moving averages and most studies fall under the category of
What Direction? After that there are various momentum indicators, some based off second derivatives of moving averages. These would be under the category of
How Fast? Lastly there are various overbought/oversold studies, stochastic indicators, mean reversion tools. These are under the category
Too Fast? I would guess that 80-90% of technical analysis tools answer the following questions about asset price movement: What Direction, How Fast, or Too Fast?
Most traders know that going against a trend is a bad idea. Dennis Gartman is well known for saying he wants to buy things that start in the bottom left of his screen and go to the upper right and sell things that start in the upper left and go the bottom right. In the NBA example from earlier this is like first finding which team won more the previous season. After determining the trend there is other information that can be added. Is it losing momentum indicating a change of trend? Has it gone up to fast and is a pullback coming? These are like adding the halftime score.
The problem with technical analysis that most traders add too much information. They take it to the point of trying to figure out what brand of shoes the players of the NBA teams are wearing in hopes that this will give them a slight edge. Since the important information is already accounted for, the rest of the information is of less and less value. I believe the proper economics term would be "decreasing marginal utility". Eventually it just becomes noise that distracts from the information that is actually of value. A quick look at a site like
Chart.ly provides a slew of examples
http://chart.ly/jc3l9vj http://chart.ly/y2vpt2u http://chart.ly/7j8ytcj
The problem with researching this is that there aren't simple testable situations. It's easy to look at last years NBA performance and see the probability that this can predict the final outcome of a game. In trading it's harder. How far back should the historical data go? What timeframe is being traded? What constitutes a win? etc... Additionally most traders aren't using technical indicators as part of a mechanical trading system. Many traders might just use it as a piece of additional information when making a final decision so it's nearly impossible to test the empirical effectiveness of these tools.
I think technical studies can provide a cleaner and more objective view of price that reduces the noise of day to day or minute to minute fluctuations. It's a tool that can provide important pieces of information that can inform an investment decision or trade. But like all information once you go beyond what is needed and try to accumulate a vast amount of additional information, it's value fades. The noise of additional information clouds out the signal.
I think the best thing to do is find a small number of indicators. Pick one that identifies a trend (my favorite is a
volatility adjusted moving average). Pick something that gives you a stop (ATR trailing stop works fine). What about How Fast or Too Fast? Eyeball it. Learn those studies. Learn how they react under different circumstances. Find when they work and more importantly when the don't work. Over time they become your last year's NBA records. They're the pieces of information that help make the best decision. Don't worry about everything else. Just take the best.
