Economics and Finance
This paper seeks to analyze the information ratio differences between long/short hedge funds over the past two decades using the Treynor-Black model. The Treynor-Black model is a method to derive an optimal portfolio allocation across safe and risky assets, based off of expected alphas of active investments and the unsystematic volatility that can be attributed to each given security. We first developed and implemented a model to forecast information ratios on a database of long/short hedge funds. With the predicted information ratios, we calculated out-of-sample allocation weights from a Treynor-Black active portfolio model. These weights were then tested in a long/short format against a Naive model that invests equally in all hedge funds. By subtracting the Naive weights from the Treynor-Black weight recommendations, we were able to test the efficacy of the Treynor-Black model under performance-neutral circumstances.
We found that the Treynor-Black model outperforms in a market that is trending upwards, such as 2017. In a market with a correction, as seen in December 2018, the Treynor-Black model performs in-line with the Naive, generating minimum excess return but taking on no additional risk. Following a market correction into another upwards market (seen in 2019), the Treynor-Black model is not nearly as effective. Due to the importance of the previous year's information ratio, the recommended allocations expected a continuation of market risk and overcorrected. We conclude that information ratio predictions combined with the Treynor-Black model can help generate alpha in a bull market, while taking on average downside risk in a turbulent market, instead of undue downside exposure as seen in some funds.
This item is protected by copyright law (Title 17, U.S. Code) and may only be used for noncommercial, educational, and scholarly purposes.
Business | Finance and Financial Management | Portfolio and Security Analysis
Pannu, Gurkamal S., "Hedge Fund Performance with the Treynor-Black Model" (2021). Honors Theses. 330.