Text based hierarchical risk parity (TBHRP)

We introduce an alternative to both MVO and hierarchical risk parity (HRP) by proposing a method known as text-based hierarchical risk parity (TB-HRP). This innovative approach leverages financial text data, specifically from 10-K filings, to capture complex economic relationships, reduce dependence on historical prices, and adapt dynamically to the changing nature of businesses. Unlike traditional techniques that rely on past returns, TB-HRP offers several advantages, including the ability to reflect operational similarities, reduce noise in portfolio optimization, reveal unexpected relationships between companies, and provide a forward-looking perspective on risk factors. Using 10-K statements of firms in the S&P 500, we apply the framework to construct a distance matrix and follow the approach similar to De Prado (J Portfolio Manag 42(4):59-69, 2016) for portfolio formation. TB-HRP can be used to enhance diversification, improve robustness, and dynamically adapt to market changes, making it a valuable tool for portfolio managers seeking to overcome the limitations of traditional optimization techniques.