Beta estimation

The sensitivity of a security towards a factor can be estimated from past price movements. We will estimate the beta from the one-factor index model. First, we show the process of collecting and synchronizing data from different sources and then present the simple beta estimation method and, at last, a linear regression model is built.

Data selection

We download the time series of the price of a given stock, for example Google, and the time series of the price of the market index, the S&P 500, from June 1^st 2009 to June 1^st 2013 from Quandl, as discussed in the second chapter:

> library(Quandl)
> Quandl.auth("yourauthenticationtoken")
> G <- Quandl('GOOG/NASDAQ_GOOG',
+   start_date = '2009-06-01', end_date = '2013-06-01')

The resulting G is a variable containing 6 variables, from which we only need the Close values:

> str(G)
'data.frame':     1018 obs. of  6 variables:
 $ Date  : Date, format: "2009-06-01" "2009-06-02" ...
 $ Open  : num  419 426 426 435...