Idiosyncratic Noise Filtered (INF) Beta

Reliably estimating the beta of an individual stock has proved elusive. Individual stock price movements are a function of market/systematic movements and company specific/idiosyncratic movements. Since traditional beta estimation calculations make no attempt to minimize the impact of the idiosyncratic movements the beta regression coefficient will be fooled to best fit both the desired market based movements as well as the undesired idiosyncratic movements. If the idiosyncratic movements are not perfectly uncorrelated with market movements the regression beta will be inaccurate.We have developed a beta estimation model that appears to all but eliminate the idiosyncratic noise and statistical chance with other calculations of beta. Once we substantially improve individual stock beta estimates then we may not have to always rely on industry averages. Moreover even if we do rely on industry averages the industry averages will be more accurate and will require a far smaller sample