Honey, I Shrunk the Trend-Following

For most of the 21st century, 60/40 has been the hallmark of multi-asset investing, driven by the twin tailwinds of rising prices and negative correlation.

Trend-Following with Long-Only Multi-Asset Portfolio

We are cognisant that a large proportion of investors’ portfolios typically comprise a long-only multi-asset (LOMA) allocation. The most common example is the 60/40 portfolio: 60% global equities and 40% global bonds, notionally allocated. There may be regional or slight differences in the ratio of stocks to bonds, but for the scope of this paper, we will focus on a global version.

For most of the 21^st century, 60/40 has been the hallmark of multi-asset investing, driven by the twin tailwinds of rising bond and equity prices, while being negatively correlated with each other. However, as we illustrate in Figure 1, an excerpt from a previous paper, the correlation between bonds and equities has recently trickled into positive territory, a reversion to what was seemingly the status quo for most periods before the turn of the century.

It may be that traditional portfolios need additional sources of diversification more than they have done in this century.

Thus, it may be that traditional portfolios, such as 60/40, need additional sources of diversification more than they have done in this century. Trend-following, we believe, fits the bill nicely, but how much do investors need?

In Figure 2 we analyse the optimal combination, as defined by Sharpe ratio, of trend-following, proxied by the BTOP50 Index, and 60/40, proxied by 60% MSCI World USD Hedged Index and 40% Bloomberg Global Aggregate USD Hedged Index, through time. The different points on the x-axis of the chart refer to the different lookback windows used when ascertaining the optimal combination but all running through to December 2023. For example, the point in January 1995 computes the optimal combination using a start date of January 1995 up to December 2023, whereas a point in January 2015 determines the optimal combination using data from that point up until December 2023.

Trend-following’s convexity means that this tracking error to 60/40 is generally at its greatest when we need it most, in other words, when 60/40 is at its worst.

Admittedly, an investor or allocator looking at the aforementioned optimal figures may be gawking at the sizes of the proposed trend-following allocations. And there is some justification in that reaction.

While trend-following is an active, absolute return strategy and therefore not managed to a benchmark, 60/40 or equities are the industry gold standard for performance evaluation. There is some merit to this. Equities, and by association, 60/40, have been one of the best performing asset classes of the last 20 years, earning it the TINA – ‘There is No Alternative’ – title. Therefore, the risk of underperforming (i.e., downside tracking error) equities or a 60/40 for a sustained period invites challenging questions from clients and investment committees and is therefore a real-world constraint to consider when designing asset allocations.

However, trend-following’s convexity means that this tracking error to 60/40 is generally at its greatest when we need it most, in other words, when 60/40 is at its worst. In Figure 4, we plot the rolling 12-month tracking error of a 100% trend-following portfolio and the Sharpe-optimal combination (40% Trend and 60% 60/40) to 60/40. The spikes in the tracking error curves are contemporaneous to drawdowns for 60/40, while the curves taper towards their lows when 60/40 rallies.

The cash efficiency of instruments traded by trend-following strategies means that little additional cash is needed to fund an allocation.

The point of this paper is not to say investors should significantly down weight the LOMA component of their portfolio and replace it with trend-following. Equities and bonds have accrued significant historical gains, and justifiably make up a meaningful portion of strategic asset allocations. As discussed in ‘Trend-Following: If it Moves, Monetise It!’, the cash efficiency of instruments traded by trend-following strategies means that little additional cash is needed to fund an allocation.

Let us take the following example. Allocating 90% of cash to the LOMA and 10% to trend-following, with the latter levered up by 4x, leads to roughly 40% trend-following, 90% LOMA, or 30% trend-following and 70% LOMA in normalised notional allocation terms. See Solution 1 in Figure 5. But this is just one of many implementation options. Another approach, that preserves 100% of the LOMA, would be to replicate the LOMA using a swap or futures, with the cash-efficiency creating excess cash to allocate to the trend-following component – Solution 2 in Figure 5. Both options also facilitate customisation in terms of the amount of leverage desired.

Aside from its desirable crisis alpha and riskmanagement properties, trend-following generates attractive longterm risk-adjusted returns while being diversified to traditional assets.

Real-world constraints often predicate that trend-following may not get the allocations our two measures of optimal, namely Sharpe ratio and drawdown mitigation, might suggest. And we are not advocating that it should either. Instead, we hope the takeaway of this paper is that investors and allocators give trend-following serious consideration for inclusion in portfolios in more meaningful quantities than it historically has been. Aside from its desirable crisis alpha and risk-management properties, it generates attractive long-term risk-adjusted returns while being diversified to traditional assets. Cash efficiency also means that it need not precipitate abandoning a traditional LOMA approach to accommodate more trend-following, but instead combined in a robust framework that uses leverage to monetise the diversification.