Global Macro Hedge Funds: Man Versus Machine
Category: Investment Research
By Jeff Egizi, CFA, Senior Research Associate
November was an unusually poor month for quantitative/systematic (computer-traded) macro strategies, with many reputable names in the space down 5 to 9%. Conversely, it was an unusually strong month for discretionary (human-traded) macro strategies, with funds in Athena’s macro portfolio up 3 to 7%. By taking a closer look at this case study, we can better understand what to expect from both strategy types.
1. Quant funds, especially trend followers, are vulnerable to sudden changes in multi-year trends.
• Trend followers, often referred to as CTAs, position for a continuation of historical price action. As a result, CTAs had been long bonds for years, a trade that has been reversing for the past few months. The mechanical trend following algorithms that most CTAs use are too slow to adapt to a sudden shift like the one we’ve seen in October and especially November. Discretionary macro funds are under no such constraints and can (1) quickly reposition when they see conditions change, or (2) already may have forecasted the change based on a fundamental view. November’s top-performing discretionary macro funds tended to be positioned for higher rates and a stronger US dollar based on expectations of rising inflation and a Fed rate hike in December, while CTAs were still fighting yesterday’s war.
2. Fundamental systematic macro funds, another breed of quant which is a systematic version of the fundamental discretionary manager (think discretionary macro fund traded by a robot), are vulnerable when prices move sharply to discount future fundamental changes which have not yet begun to occur.
• These systematic macro strategies assume the world will stay the same, look for large deviations between prices and fundamentals, and position for mean reversion. When those deviations continue to widen because markets are reacting to something not present in the current data (such as a shift towards fiscal expansion), systematic macro funds are caught swimming against the tide. Conversely, discretionary macro managers can interpret the fundamental implications of politics and other exogenous events and position accordingly. Markets such as rates, the US dollar, copper, and small cap have moved drastically to discount the Trump administration, when in reality nothing has occurred to impact current fundamentals. Certain systematic macro funds were starting to take the other side of the rates sell off in October due to the ongoing stagnation in the US economy, which caused losses when bonds legged even lower on the Trump election.
3. Most quant funds use some form of optimization for portfolio construction, and therefore are vulnerable when asset covariances change suddenly and significantly.
• This was more of an issue in October when equities and bonds became positively correlated, but serves as a reminder that mean-variance optimization can be problematic. Portfolio optimizers which maximize return/risk are making assumptions about the relationships between different asset classes, usually based on past data. For example, an optimizer that assumes equities and bonds are negatively correlated will “think” it can allocate to more of each and maintain target risk levels, because the price movements should offset one another. However, when markets start to worry about inflation, bond and equity correlations can flip to positive. Over the long-term there have been prolonged periods of positive equity-bond correlation, but current optimization models are trained in the negative equity-bond correlation status quo. If this regime changes, the result will likely be portfolios being more volatile than anticipated, leading to deleveraging by quantitative funds and exacerbating market volatility.
So why ever allocate to quant over discretionary? The general rule of thumb is that quant funds will make a lot of trades, all of which have a slightly better than even chance of being profitable, without the cognitive biases and limitations that come into play with human PMs. Also, given the growth in data and computing power, the discretionary macro information advantage of 'knowing a guy in politics' might not be as strong as it once was going forward.
In addition to the “equity diversifier” argument we have made about systematic funds in the past, there is also a strategy diversification argument that bolsters the case for quant funds. Specifically with respect to fundamental systematic macro funds, if the world ends up staying the same (see example in point 2 above), fundamental systematic macro funds will capitalize on the reversion, while discretionary managers will give back profits. Therefore, having both fund types in the portfolio can dampen volatility.
In conclusion, we believe that neither a human nor a computer is the inherently superior macro portfolio manager. Quantitative funds have historically done a better job diversifying long-only equity exposure, and may be able to capitalize on advances in data and computing power in the future. However, skilled traders can position for and protect against events that are not easily observable in the data. Therefore, there is room for both strategy types in a balanced macro portfolio.