Trend-following and the RORO (risk-on/risk-off) regime

The talk of risk-on/risk-off trading has been the key uncertainty driving markets for the last few years. It could be one of the chief reasons for why longer-term trend-followers have recently had a hard time making money and why fundamentalists have been unable to exploit economic imbalances across countries in the FX markets.  It may not be appropriate to call RORO the period’s defining market paradigm or regime, but it is clear that RORO behavior was not strongly present prior to the financial crisis.

There is no agreed definition of what is a risk-on/risk-off trading behavioral regime. However, we can identify environments that are more subject to this type of risk, those with increasing correlation across markets.^[1] We have informally discussed these environments before through comparing correlation heat maps which provide a static snapshot of changes in correlation. In a RORO regime, markets move together in a manner that is reinforcing; however, this is without clarity on the direction of causality.^[2] Sensitivity to a single macro factor will create higher correlation across markets. More highly correlated markets suggest more RORO behavior to any single market event.

Clearly, RORO switching between safe and risky assets is focused on changes in common macroeconomic expectations which cause all markets to move together. Nevertheless, it is hard to define what specific economic events will drive RORO trading. It can be a policy or data announcement or even a press conference comment by a government official. Risk aversion indices can also capture some of the factors which seem to lead to RORO trading unrelated to specific macro-events. Factors which express investor sentiment on risk aversion such as the VIX index or credit spreads are indicative of market risk perception. 

All these events cause markets to quickly adjust asset allocations in order to avoid market downturns or participate in market gains. Nonetheless, the sets of factors which drive RORO behavior are unclear. Hence, there needs to be a more precise way to isolate and research RORO trading as something unique.

How does this relate to trend-following? One of the key characteristics for longer-term trend-following is that trend are relatively smooth. The volatility around the trend is not large and there is diversification across markets. 

The diversification across markets is necessary because the success rate for most trend-following models is low. If there is more correlation across markets, there are fewer trends and the risk associated with any program of trend trades actually increases. Hence, the risk for a trend follower is higher. 

The type of volatility is important because trend-followers are actually long long-term volatility and short short-term volatility. The trend-follower likes long-term volatility because this supplies or represents market range. If there is no range over the long-run, there is no opportunity for the trend to establish itself and have a chance for accrued profits.  With RORO trading, there is less chance for long-term trends because the market moves between holding risky assets and safe assets. The switching causes significant short-term price reversals which create noise for the trend-follower. It is the short-term volatility which is disliked by many longer-term trend-followers. Short-term volatility will increase the amount of noise which will reduce the change of finding trends. The noise will be unrelated to the supply and demand of a specific market dislocation. The trend has to be stronger to provide a clear signal around the noise. 

Since trend-followers often have stop loss provisions with their models, short-term volatility increases the chance that stops will be hit. A good trend trade could end because of the stop. Hence, short-term increases in volatility actually reduce the success of trend-followers. A RORO environment will increase this short-term volatility. This will be bad for trend-followers. It should not be surprising that trend-followers have underperformed. They will have to wait for larger moves with one directional volatility that can be exploited. 

If RORO lasts for a long period, that is, risk-on or risk-off lasts for an extended period, there will a chance for strong profits. Periods of strong managed futures performance have been associated with extended risk-off behavior.  

---

[1] Research on changing correlation across asset classes has been undertaken by both academics and major banks. The focus of this work has been on how correlations across a broad set of assets change through time.  In particular, a large correlation matrix of assets can be adopted to measure joint correlation for the entire set. Statistical analysis, using principal components, can measure the extent that a single factor or component is associated with the correlation change.  This type of work has shown that correlations across asset classes are on the rise since the financial crisis of 2008. The higher correlations across all assets find markets more susceptible to a single factor driving performance.  

[2] The reasons for the higher correlation across markets are manifold. The globalization of financial markets with limits on capital controls allows the free flow of funds to impact multiple markets over short time periods. The focus on asset managers to tactically move money across asset classes and manage the entire portfolio has also caused a stronger link across markets. Research has also found that correlations will increase during market downturns and recessions which will explain the jump in correlations during the Great Recession. Research has also found correlation of asset classes increases when volatility increases which we suspect causes more risk averse behavior. The RORO regime may be with us for some time.