Google (NASDAQ:GOOG) has been criticized for its heavy investment in data centers. Some intelligent critics argue that the margins on data centers won't provide the level of return investors expect from Google. Google's core business is in "Search", and that business appears much more profitable. Does Google know something we don't?
Google is known for innovation, and they may be leading the way again. Consider these three pieces of news together:
- Google invests heavily in data centers. According to their summary, capital expenditures were $2.65 billion in Q2 2014.
- Google Fiber is introducing Google as an Internet Service Provider. The prices they are offering are far more competitive than necessary in an oligopoly.
- An analysis of search terms may be predictive of stock market crashes.
The article on stock market crashes, ran by Daily Mail and linked above, quotes Tobias Preis saying: "The strength of this relationship, using this very simple weekly trading strategy, has diminished in recent years". If we believe that markets are semi-strong form efficient, we would expect that publicly available trading data has no predictive power. However, information that is not publicly available could still be used to develop positive risk adjusted returns. Under "Mosaic Theory" analysts could combine non-material non-public information with material public information and the resulting analysis may be considered legitimate. The Mosaic Theory defense doesn't always hold, as Raj Rajaratnam was sentenced to prison years ago.
The suggestion that the predictive power in a simple test is weakening does not mean the strategy did not work. If it does work, hedge funds applying the technique and securing the profits would eventually force the price of securities to more closely match the projected prices. If that is the case, eventually these profits would go to whatever entity could most rapidly acquire the information and modify their holdings. In order to acquire the profits, they would need to trade first. By trading first, the timing of the stock price movements would change. Consistency is the backbone of any statistical model, so a change in patterns as an actor absorbed and used the new information could cause the model to appear less valid.
Theoretically, if Google planned on using this information, where would they want to store and analyze the data? It seems that if they had this goal, Google would want to store the information in their own data centers to reduce the risk of other parties gaining access to it. That would be a major incentive to invest in data centers, even if the data centers themselves appeared to be a very low margin area.
Theoretically, if Google was intending to analyze this data and use it, they might want to ensure no other party could access to it. As long as the searches are going through Google, there should be only a few entities that could figure out what was searched: Google, the person searching, anyone with access to that person's computer history, and possibly the ISP that was transmitting the data. Those ISP's might not have a reason to analyze your searches yet, but Google also has no reason not to cut them out. Even if such an analysis by an ISP was considered illegal, it wouldn't be the first time for an ISP to violate the law regarding user activity.
If Google did decide to analyze search terms and create statistical models to flag potential searches that might precede a drop in price, would it be defensible under Mosaic Theory? It seems that it might be. The information about a single search could be considered non-material nonpublic information. If the searches are going through Google's search engines and originating from a computer using Google for internet access, no other company in the United States would be able to duplicate it.
It seems logical that there would be a spike in searches prior to an increase in volume. If those searches are matched with other key words, it could indicate if the investor was seeking to confirm positive or negative news. Even if the trading was automated, Google may only need the search to go through their servers. If Google knows what is being searched, by person or script, it may be able to predict trading patterns. They don't need to be perfect. They only need a model that can predict movements more cost effectively than the competition.
Disclosure: The author has no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. The author wrote this article themselves, and it expresses their own opinions. The author is not receiving compensation for it (other than from Seeking Alpha). The author has no business relationship with any company whose stock is mentioned in this article.
Additional Disclosure: All information in this article is the opinion of the author. Sources are deemed reliable but not guaranteed. Forward looking statements contain projections that are uncertain. The analyst is not a lawyer and does not provide legal advice on the merits of Mosaic Theory as a defense. Before investing in any security, you should speak with a financial advisor that understands your risk and return objectives. The author does not directly own Google, but does have long positions in index or mutual funds with long exposure to large cap U.S. corporations.