Several financial institutions have proprietary trading algorithms that make them fortunes on a day-to-day basis. You don’t have to have a PhD in statistics to create your own algorithms. Many hedge-funds use a powerful modeling technique called regression analysis to make their investment decisions. They are basically betting on a stacked deck.
Successful financial institutions use large data sets to predict movements in stock prices. A little known tool that compiles information on search queries is Google Trends. Using this large data set provided by Google traders can act on this information to make profits. The concept is simple; there is high correlation in company’s Google’s queries and sales. For example, in the retail business where online sales are a large source of income, it seems logical that search queries for Abercrombie and Fitch are correlated with sales. This strategy is transferable to several industries such as; automobile, insurance, apparel, technology, etc.
It is crucial to understand the concept of regression analysis in order to capitalize on this information and create profitable opportunities. There are plenty of websites that do a thorough job of explaining regression. You can find my brief explanation below, with the help Wikipedia.
Regression analysis is widely used for prediction and forecasting. In statistics, regression analysis includes any techniques for modeling and analyzing several variables, when the focus is on the relationship between a dependent variable and one or more independent variables.
Regression models involve the following variables:
The unknown parameters denoted as β
The independent variables X.
The dependent variable, Y.
A regression model relates Y to a function of X and β.
Y = f(X,β)
In the last case, the regression analysis provides the tools for:
1. Finding a solution for unknown parameters β that will, for example, minimize the distance between the measured and predicted values of the dependent variable Y (also known as method of least squares).
2. Under certain statistical assumptions, the regression analysis uses the surplus of information to provide statistical information about the unknown parameters β and predicted values of the dependent variable Y.
In linear regression, the model specification is that the dependent variable, yi is a linear combination of the parameters (but need not be linear in the independent variables). For example, in simple linear regression for modeling n data points there is one independent variable: xi, and two parameters, β0 and β1:
Straight line: yi= β0 + β1x1 + β2x2 + … βnxn + є
є accounts for error in the regression.
Data should be added and removed to provide the best possible regression prediction. Also, there are several forms of regressions that include non-linear and logarithmic. You can use these to generate a more accurate best fit line.
Hopefully I can walk you through a general example to clarify any misunderstandings that you may have at this point.
Let’s choose the automobile industry for this example. I use Ford’s profits (in million dollars) as my dependent variable (y) and Ford’s queries (x1), Toyota’s queries (x2), Honda’s queries (x3), and GM’s queries (x4) are independent variables. These queries are extracted on a weekly (you can choose whichever timeframe you like) basis. Using the OLS regression function in MS Excel I highlight the dependent variable (y) column and independent variable columns (x). The regression should spit out two key pieces of information, your best fit line and R2 value.
Your best fit line should look something like:
Y(x)= 1.25 + 4.23x1 – 1.14x2 – 1.03x3 + 0.54x4 + є
The reason why GM and Ford have positive values is because queries of these companies positively correlate with Fords profits. While Ford’s positive value seems immediately intuitive, GM’s may not. However, after putting some thought into why GM’s queries positively impact Ford’s profits, this can translate into more consumers purchasing U.S manufactured cars. Honda and Toyota have negative coefficients because obviously the companies are taking away market share from Ford, and thus lowering profits.
This equation that we observe below can be used to predict Ford’s profit if the best fit line is 100% accurate. Unfortunately, it is not entirely accurate for several reasons that I will not get into. We can get a better model by inserting or removing certain data sets. This will increase our R2 value. The closer this value is to 1 the more accurate our best fit line is to the data we have used to forecast. If you are trying to implement this strategy I recommend that you try to improve the R2 value by adding and removing different automobile manufactures in order for you to get the value closest to 1.
The beauty of Google Trends is that the queries are complied into neat searchable categories. Also, the amount of data is mindboggling.
This confirms that fact that using Google Trends to formula a best fit line using linear regression we can in fact predict the present performance of sales which correlates with share price.
Investors often make investments based on earnings numbers that are indicative of sales however, earnings numbers are backward looking. Meaning, the market has already priced in this information into the stock. This linear regression model can help you trade ahead of earnings news, allowing you to make a nice profit.
If you set up the model in MS Excel once, it can be duplicated and simply spit out relevant information on a per sector basis. All you need to do is search Google Trends and copy paste the data.
Can Google Predict the Future?
Posting Permissions 