Infantino's 2010 PCA Model

Overview

This model, from a large group of stocks e.g., S&P 500, computes, using ​Principal Component Analysis, the common factors that drive stock returns and filters out noises. We can then estimate the short term expected returns for each individual stock, hence a trading decision.

Moreover, the model proposes a way to use intraday volatility to estimate which of the two regimes we are in -

• mean-reversion, or
• momentum

Combining the estimated expected returns with the regime indicator, we do mean-reversion when the intraday volatility is small, and trend following when the intraday volatility is big.

References

• ​http://dspace.mit.edu/bitstream/handle/1721.1/59122/658860705.pdf

Code

• http://www.numericalmethod.com/trac/numericalmethod/browser/algoquant/core/src/main/java/com/numericalmethod/algoquant/model/infantino2010