Trend-followers are not alike - The case of turbulence

 

The concept of turbulence has been well-developed by Kritzman and Turkington using a distance function. You want to look at the joint behavior across assets in manner that accounts for both volatility and correlation. From this distance function we can measure the correlation surprise, the impact of changes in the correlation across assets as well as shocks to volatility. The use of distance functions provides simple means of accounting for both volatility and correlation in a single measure of disruption. 

Using the turbulence distance function, events can be decomposed into magnitude or correlation blind measures when the off-diagonals are set to zero and correlation surprises which is the ratio of turbulence versus magnitude measures, or volatility shocks and correlation shocks from the turbulence measure.

The turbulence function has been recently used in a paper to measure disruption across a set of large trend-followers. Periods of high turbulence across managers can be matched with market events to determine when diversification across managers is most needed. See "Quantifying Turbulence in Trend-following". 

The results suggest that turbulence shocks are associated with specific market events. Some of these events raise volatility across all managers, magnitude shocks while others will be associated with correlation shocks which are likely associated with differences in portfolio compositions across managers. Investors need to be aware of these shocks to build better portfolios. High turbulence will in general show low returns.

Trending the trendfollowers - Can it be done?

 

Can you trend-follow the trend-follower? This is an important question give the strong performance of trend-followers in the last year. Put another way, is it too late to allocate to the trend manager? Should you wait for the next trend drawdown? 

We don't have a solid answer but would like to present some ideas for consideration. 

1.    Asset return behavior is different from strategy return behavior. Asset prices can go through long rising or falling periods and there can be an assessment of fair valuer for an asset which will impact trend behavior. Trend-following as a strategy does not have a fair value and positions can change quickly across many markets, so asset and strategy behavior cannot be compared.

2.    Trend-following attempts to exploit any autocorrelation within the assets it trades. If this is done effectively, there is will little autocorrelation in the strategy itself. See "Can you trend follow trend-following?". 

3.    Trend-following is regime dependent with returns often clustered for short-periods. Some call this the crisis alpha effect. Unfortunately, it is hard to determine when there will be a crisis and when a give crisis will end. 

4.    Diversification across asset classes makes it harder to determine which assets will be the driver of return. 

5.    Trends will often last longer than expected. Hence, it cannot be said that trend performance will reverse.  However, there are long-term Sharpe ratios that can be a guide for performance.

6.    A trend that ends can just mean a reversal in a position from long to short. The risk is the difference between the maximum profit and the time until the position is reversed.

7.    Buying into drawdowns is not based on mean reversion given positions can be long or short. There is no mean reversion based on the assets bought. A drawdown could mean that old positions have been cleared within the strategy. A drawdown may also be associated with deleveraging which reduces return potential. Mean reversion during a drawdown is based on skill assessment or whether the manager just faced bad luck during a drawdown.

8.     Trend-following is impacted by volatility - both the level and change. Higher volatility is good but changes in volatility can hurt performance.

The timing trend-following is not easy and should be done with care; nevertheless, investor should account for the market regime and the current Sharpe ratio versus the long-term Sharpe of the manager and the overall strategy. The Sharpe ratios will mean-revert.

Polycrises - multiple crises are a reality

Policymakers and investors have increased their awareness of systemic risks and crisis events. Policymakers have developed macro prudential policies to limit the impact of systemic risks. Investors have spent focused time and effort on tail risks and how to mitigate large losses from a market crisis. However, most of the focus has been on a single crisis or shock. Less time has been spent on thinking through the impact of multiple events that can all cause harms.  If crises occur at the same time, a small crisis may turn into a catastrophic event. These correlated events would be a polycrisis. See "What Is a Global Polycrisis? And how is it different from a systemic risk?"

From the Cascade Institute - "A global polycrisis occurs when crises in multiple global systems become causally entangled in ways that significantly degrade humanity’s prospects. These interacting crises produce harms greater than the sum of those the crises would produce in isolation, were their host systems not so deeply interconnected."

So, what is the possible polycrisis we are currently facing? Let's just list some of the intersecting issues: Ukraine-Russia War, the pandemic residuals, the China-Taiwan-US situation, the deglobalization issue, the climate crisis, the inflation crisis, the excess money and debt crises, a commodity crisis, and an overvalued stock and bond market. 

We will do walk through all these issues other than to say that the solution of one may conflict with another. Current energy security conflicts with climate change. The geopolitical issues conflict with inflation. The inflation problem conflicts with an overvalued market problem. All these crises are interconnected and complex. Policymakers will not be able to solve all. Solving one may make others worse. Each may change in importance over time, so it is difficult to forecast market behavior. A polycrisis adds to complexity and uncertainty. We live in a VUCA (Volatility, Uncertainty, Complexity and Ambiguity) world.    

Survivorship bias - Need to know what is missing from the data

 

There is the old story concerning Abraham Wald, the great statistician, and measuring the right thing in any statistics problem. The Air Force in WWII wanted to determine how they should add armor to their bombers to ensure they would survive given heavy losses from its daylight bombing. The air force staff gathered statistics on all the bombers that came back after missions and looked at the probability of certain areas of the plane being hit with flak or bullets. The idea was to add armor to those areas most often hit with enemy flak.

They proudly gave their extensive evidence to Professor Wald and asked him to validate their thinking on where to put the most armor. He responded in a very simple way by saying that armor should be placed where there was no record of damage. This was just the opposite of what was expected by the other statisticians. His answer was simple and profound, "The bombers hit in those places never came back." The data analysts could count the surviving bombers' damage. There is no evidence for the bombers that crashed.

Ask what data are counting and then ask what the data are not counting. Only survivors are counted and included in databases. You also want to know what got away from the analysis.