Covariance Matrix of Trends and Risk Premia in Portfolio Allocation
Optimal portfolio allocation is a hard problem to solve in finance. Determining the best way to allocate assets and strategies across different investment vehicles in order to achieve the best possible risk-adjusted return is a complex task that has been studied by mathematicians and financial experts for many years. Despite all of the research that has been done, there is still no definitive answer to this question. There are a number of methods available for finding an optimal portfolio, and each has its own strengths and weaknesses. Reference [1] introduced a new method for allocating trend following strategies that utilized not only the covariance matrix of returns but also the covariance matrix of trends and risk premia.
We derive a theoretical setting to yield implementable solutions of the allocation problem of trend following portfolios. The main formula of the paper describes the optimal portfolio as depending on the covariance matrix of returns, the covariance matrix of trends and the risk premia.
We implement the formula to gauge the performance of five well established portfolios (Agnostic Risk Parity, Markowitz, Equally Weighted, Risk Parity and Trend on Risk Parity), using daily data from futures markets of 24 stock indexes, 14 bonds indexes and 9 FX, from 1985 to 2020.
Our main empirical finding is the optimal combination of the three best portfolios produces a Sharpe ratio of 1:37, with their respective optimal weights of 19.5% (ARP), 51% (RP), and 30% (ToRP) which combines both traditional and alternative approach. Consistent with related recent literature, we confirm that RP portfolio, which is a proxy of the traditional and well diversified portfolio is a important driver of performance. Furthermore, we show that the combination between ARP and ToRP is the best solution in term of Sharpe ratio for the trend following approach and the alternative benchmark as they tend to minimize the correlation among assets.
In the context of a portfolio optimization problem, the article solved for the optimal allocation amongst a set of trend following strategies. It utilized the covariance matrix of returns, trends, and risk premia in its optimization algorithm. The allocation scheme combined both traditional and alternative approaches, offering a better Sharpe ratio than each of the previous methods individually. We note, however, that it still depends on historical data and has not been tested for robustness.
References
[1] Sébastien Valeyre, Optimal trend following portfolios, https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4908749
Great post. This might help with portfolio allocation. The challenge in portfolio allocation is that correlation is often a living thing. Many assets are perfectly uncorrelated until a stock market crash hits, and exactly when you need the uncorrelation, they become correlated. This is not an argument against portfolio diversification, just highlighting how complex that topic is. Maybe research like this helps.
Another great paper!! Thanks for sharing!!