Good day. I am looking to understand how the portfolio construction process is actually done in the industry. Now, I do not know if there are too many resources on how things are currently being done (I would imagine not many are sharing their secrets), but I would like to get a practical understanding (i.e. implementation) of the current methods and techniques applied in the industry. I would be grateful if I could be directed to some resources. A cookbook-type resource with code examples would be great, but theoretical resources are equally appreciated
If my understanding is correct, the portfolio construction process entails the forecasting/estimation of moments (expected returns, covariance matrix, maybe higher moments), an optimization procedure to obtain portfolio weights, and then backtesting of the strategy. Please let me know if I am entirely wrong, clueless about this, or missing a crucial step.
Among the specific questions I have:
What are the current techniques used to obtain forecasts of moments? Are factor models used in practice to obtain, for example, the expected return for the next period (t+1)? or is this something that no one does?
For the optimization procedure, is maximizing the expected return subject to constraints a sensible way to approach the problem? if so, What constraints are often used? both for controlling for risk (portfolio variance,cvar, constraints in portfolio betas, or other measures currently used) and other elements such as turnover
In practice, is single-period optimization preferred over the multi-period? I would imagine that if multi-period is used, one would keep only the first period's result and repeat the multi-period optimization the next day ( or the next time the portfolio is being adjusted) to incorporate new information but maybe I am wrong about this
Thanks in advance for taking the time to read this.
