Descripción
A new method for portfolio optimization using so-called "Hvass Diversification", which is extremely fast to compute and very robust to estimation errors in the correlation matrix. We also show why Markowitz "mean-variance" optimization is horrible.
Papers:
https://ssrn.com/abstract=3942552 (Full paper)
https://ssrn.com/abstract=4009041 (Only diversification algorithm)
https://github.com/Hvass-Labs/Finance-Papers (Backup)
Data and computer code for the paper:
https://github.com/Hvass-Labs/FinanceOps
Python package and tutorial:
https://github.com/Hvass-Labs/InvestOps
https://github.com/Hvass-Labs/InvestOps-Tutorials
00:00:00 Introduction
00:00:32 Critique of the political left
00:13:34 Critique of academia
00:18:25 ** Main content begins **
00:20:11 Mean-Variance Optimization (Section 2)
00:20:33 Variance is NOT Risk! (Section 3)
00:34:22 Naive Forecasting (Section 4)
00:37:21 Conditional Forecasting (Section 5)
00:41:39 Random Walks (Section 6)
00:44:19 Filtering Methods (Section 7)
00:47:15 Diversification Method (Section 8)
01:07:30 Test Settings (Section 9)
01:13:43 Test A - Full Data Period (Section 10)
01:26:15 Test B - Data Until 2010 (Section 11)
01:26:45 Test C - Data From 2010 (Section 12)
01:27:28 Test D - Noisy Returns (Section 13)
01:28:48 Test E - Noisy Correlations (Section 14)
01:31:35 Test F - Noisy Returns & Corr. (Section 15)
01:32:15 Test G - Parameter Tuning (Section 16)
01:36:15 Computer Code
Cartoon by Scott Adams (https://www.dilbert.com)