Bringing an End to Errors: Part 4

  • Thursday, September 28, 2017

This is the 4th part of a 6 part series on how cutting-edge text analysis technology is helping investors mitigate the detrimental effects cognitive biases can have on their decision-making.

Solutions for Selection Bias

Although herd mentality can be detrimental to effective financial decision making, there is a significant element of investing that has everything to do with measuring the opinion of the herd. Prices are, after all, a quantitative measure of public opinion, and, if an investor can anticipate shifts in opinion before they are captured in price, that investor stands to make quite a bit of money. It is here that selection bias can cause real problems.

A victim of selection bias relies on “samples that are not representative of the whole population they’re studying.” When it comes to assessing the sentiment of market actors, this can be an easy trap to fall into. The opinions of the masses are not easily accessible before they’ve affected price, and the opinions that are easily accessed, whether on Twitter or television, are not necessarily representative of the masses. Consequently, financial professionals are often stuck choosing between no data and poor data—and they often choose the latter.

Once again, emerging text analysis technology presents a solution to this problem.

Before we elaborate on this point, it must be noted that standard social media sentiment analysis is not the solution being referred to here. Methodology matters. Such approaches commonly use the social media sentiment associated with a specific asset, commonly a stock, as an analogue to the market sentiment around that asset. But, as we discussed earlier, many who won’t express their thoughts on Twitter will do so through trading. Consequently, these approaches do not represent a means of checking selection bias as they fall prey to it themselves.

Instead, the approaches that do represent a potential solution must use a better measure of sentiment. For investing applications, a well engineered textual analysis system evaluates language in light of the history of the market response similar language has garnered. Market reaction is not only the most important dimension to investors, it is also clearly “representative of the [relevant] population.” Market reaction may not be the wisdom of crowds, but it is the expression of the will of the relevant crowd—the one that drives price movement.

So, the next time you’re perusing a financial publication and your favorite analyst notes that the tone of Apple’s earnings call doesn’t bode well for the tech leviathan, stop before you short and remember: the impression of your favorite analyst is not necessarily the same as the impression of the market.

The Prattle Team

New Call-to-action