Man Institute Trend Following Cash Is King

Trend followers have long sought to expand into new markets offering differentiated, idiosyncratic return drivers. Within equities, cash equities offer an opportunity to do precisely this, opening the door to trends inaccessible through index futures alone Not all cash equity approaches are equal, however. Historically, directional sector trend following has generated positive absolute returns, but remained correlated to equity index futures trend, particularly in the tails We believe cross-sectional strategies, applied to sectors and equity styles, can offer more diversifying characteristics and serve as an effective complement to traditional trend following

Trend following in equities has historically been concentrated in a finite, but broad set of index futures – the S&P 500, Korean Kospi, MSCI EM, and Nifty, among others. This is largely borne out of the fact that they are among the most ‘trend compatible’ ways of obtaining exposure to the asset class: operationally straightforward to trade, highly liquid and capital efficient. As we demonstrated in our work on the optimal market mix, equity index futures play a crucial role in a trend-following portfolio. They are both highly diversifying (when combined with other asset classes, including bonds, commodities and currencies) and highly sensitive to large macroeconomic shifts, including monetary policy changes, inflation regimes and global risk appetite. This macro sensitivity underpins trend followers’ ability to capitalise on broad market selloffs or bull runs. Put differently, equity index futures are a key component of the convex return profile trend-following strategies can offer. Notwithstanding, trend followers have consistently sought to complement this macro risk exposure with alternative markets, driven by more idiosyncratic risk factors, in a bid to bolster diversification. Within equities, drilling down to the single name level offers one way of doing this. In what follows, we examine the ways in which cash equities can be deployed within a trend-following framework, assessing the diversification each approach may deliver relative to traditional equity index futures trend.

The typical starting point for implementing trend following on cash equities is to first group stocks into baskets, rather than applying models directly to single names. The rationale for this stems from empirical evidence presented by Jegadeesh and Titman (1993) and De Bondt and Thaler (1985), showing that individual stocks exhibit reversal effects at both short and long horizons. Compounded by idiosyncratic noise and event risk, these effects can hamper the development of profitable trends. Grouping individual cash equities into baskets, such as those based on sector or industry classifications, helps mitigate these headwinds by smoothing out stock-level noise and isolating more persistent, basket-level price movements. As such, the starting point for our analysis will be sector-level baskets, based on GICS level 3 (GICS3) definitions. One might expect directional, time-series trend-following models on sector baskets to be able to capture dispersion inaccessible from trading the index alone. During a sector rotation, like the one seen in the first quarter of this year, such a strategy could be long semiconductors and short software and services, capturing intra-index diversification that is masked at the aggregated level. While conceptually appealing, examining the realised correlation between a trend-following strategy applying univariate, time-series models on GICS3 sectors and traditional equity index futures reveals that the diversification benefit is less meaningful than expected.

Source: Man Group database. Date range: January 2005 to February 2026. As Figure 1 illustrates, the rolling 24-month correlation between equity index trend and directional, time-series sector trend following is persistently high, averaging approximately 0.81 over a 21-year sample. This likely reflects the fact that both approaches are inherently taking directional bets on equity beta or risk premia, differing only in granularity of exposure. Figure 2 examines this further, decomposing the correlation by return quintile of equity index futures trend.

Source: Man Group database. Date range: January 2005 to February 2026. Correlation is highest in both tails – and particularly so in the left tail – suggesting directional, time-series sector trend may be susceptible to the same catalysts that drive equity beta (and consequently index futures). While both strategies share positive traits, including positive long-term returns and macro sensitivity to deliver a convex return profile, this also increases the likelihood of contemporaneous reversals around market inflection points, where trend following is already most vulnerable. Clearly, this is an undesirable feature for markets that are intended to be more idiosyncratic. So how can we better monetise trend following on cash equities?

One option is to explore different ways of creating equity baskets, which we will return to shortly. Another is to reformulate portfolio construction on the same basket. Rather than taking directional bets on GICS3 sectors, we can adopt a cross-sectional, market-neutral approach. To put it another way, rather than applying univariate time-series models, we can apply a multi-variate framework based on the relative momentum of the sectors. This means being long those sectors exhibiting the strongest past performance and short the weakest, creating a portfolio with ex-ante market neutrality. As shown by the aqua blue line in Figure 3, this neutralises the implicit beta loading of directional sector trend, resulting in a fundamentally different risk profile and improved diversification properties.

Source: Man Group database. Date range: January 2005 to February 2026.

As hinted at above, we can also slice the equity universe along a different plane, using well-documented investment factors or styles. Using firm-level accounting metrics, we can form portfolios based on styles such as quality, size or value, among others, and therefore widen the opportunity set to capture a fundamentally different axis of cross-sectional dispersion. Additionally, Gupta and Kelly (2019) provide robust evidence for the alpha generated by applying time-series momentum models on equity style factors, showing that 59 of the 65 characteristic-based factors investigated exhibit positive monthly autocorrelation. Their combined factor trend portfolio also outperformed both a traditional stock momentum and industry momentum (similar to our cross-sectional sector approach presented above) strategy on a risk-adjusted basis. Beyond showcasing the efficacy of the strategy, the authors present further evidence of equity style momentum’s pervasiveness across different formation windows and regions, underscoring the robustness of the effect. In Figure 4, we construct a similar style trend-following strategy, trading approximately 13 equity styles* across each of the US, Europe, Japan and pan-Asia. Comparing this against a long-only benchmark of market-neutral styles demonstrates the additivity of using trend-following models to time returns.

Source: Man Group database, Bloomberg. Date range: April 2011 to December 2025. Equity market neutral (EMN) benchmark constructed as a composite of 35% Dow Jones Thematic Market Neutral Value Index, 15% Dow Jones Thematic Market Neutral Quality Index, 35% Dow Jones Thematic Market Neutral Momentum Index and 15% Dow Jones Thematic Low Beta Index, scaled up by a diversification benefit factor such that the benchmark achieves the same level of annualised ex-post volatility as the Style Trend strategy. The weights to each style benchmark are chosen to map as closely as possible (from the universe of style benchmarks available) to the weights of different styles within the trend strategy. * Note, the estimated 13 styles per region quoted is an aggregate figure, where different measures or formulations of a given style are grouped together. The strategy trades around 80-100 unique formulations per region, with the exact quantity varying based on the region. The efficacy of style trend following arguably stems from a fundamental market inefficiency. Equity styles are subject to persistent capital flows from various market participants (Figure 5), creating exploitable patterns driven by mechanisms such as institutional rebalancing, style rotations and risk premia harvesting. These are distinct from the speculative or fair-value price-based dynamics that drive traditional, non-synthetic markets, such as index futures. Furthermore, the market neutrality of the underlying styles further reinforces their diversifying properties relative to equity index futures trend following.

Source: Man Group database, Bloomberg. Date range: April 2011 to December 2025.

Having established the conceptual merits of multiple cross-sectional approaches, we return to the central question: how much diversification do they actually deliver? Figure 6 replicates the analysis from Figure 1, but compares the correlation between equity index futures to each of the three cash equity strategies. Both cross-sectional sector trend and style trend exhibit an average correlation of approximately 0.05 over the 21-year sample of data, peaking at a maximum of around 0.4 and a trough of approximately -0.2.

Source: Man Group database. Date range: January 2005 to February 2026. Another way of visualising the diversification is by analysing the average correlation between the cross-sectional cash equity approaches and equity index futures per each return quintile of the latter, similar to Figure 2. The below highlights that both strategies exhibit effectively zero correlation irrespective of the return regime.