There has been a story doing the rounds in the past week about researches demonstrating the efficacy of tracking key word searches to predict market direction. One news report proudly trumpeted that it was more effective than a stockbroker in making you rich, you can insert your own snarky remark about this. The use of Google trends is not a new thing – epidemiologists have found it useful in tracking influenza outbreaks.
What the researchers did was build a simple trading strategy based upon the appearance of key words. In this instance the researchers used the word debt since this seemed to test best. Using this word did seemed to substantially boost the performance of their portfolio. However, the second best performing word was colour, the third best was restaurant. This somewhat obscures what this study may or may not be telling us. If seemingly random words with no relationship to trading test well then there does seem to be a problem with the use of this as a trading system.
The interesting thing about the use of the word debt is that it seems to be postdictive, since debt was only spotted as culprit after the market had imploded and some serious navel gazing was done. This raises the question as to what word would you use to track the health of financial markets if you only know about the reasons for a financial collapse after the event. Being a Monday morning quarterback is pointless in trading. Everyone can look back and tell it was obvious – that has been the role of economists for centuries.
The researchers in explaining their results think that what we may be seeing a form of psychological insight. The use of the search term debt in the in minds of the researchers may be reflective of loss aversion. If we assume for an instance that this system works I am not so certain we are seeing loss aversion but rather cognitive dissonance. By searching for the word debt people are seeking what is known as dissonance reduction – they are searching out terms that reinforce their existing beliefs. Traders en masse do not seek out information that conflicts with their belief system – they seek out information that confirms what they already think they know.