Extension of the Quadratic Risk Optimizer on the N-Alpha Problem with CVXOPT
In the previous post, we introduced the quadratic optimization problem for portfolio returns and shared code using CVXOPT.
In this post, we extend the N-asset problem into the N-alpha problem with modifications to the constraint matrices. In the coming posts, we will release our formulaic alpha reports. We then introduce change of numeraire techniques into our market notes. Here is our current short term vision over the next few weeks; we intend to keep the portfolio management discussion on hold and finish up the discussions on stochastic calculus - the reason being it is somewhat distracting for me to keep jumping between different topical themes. After we clean up stoCalc over a few weeks, we will jump back into portfolio management at full steam. The vision is to then start including code for multivariate portfolio optimization methods into the Russian Doll, extending our quant library to a more advanced level.
Currently, our Russian Doll allows us to easily backtest optimized code for quantitative strategies, analyse return profiles and statistical significances. Once we add the portfolio optimization library, the Russian Doll will be a powerful module to run single-strat and multi-strat quant portfolios.
Preview:
Paper:
Code: