CTAs and managed futures: industry, position sizing, costs and crisis protection

In depth12 min read · 13 sources

Managed futures are funds and mandates that trade futures contracts on equity indices, bonds, currencies and commodities, both long and short. Under US law their managers are called commodity trading advisors, or CTAs for short, and most of them follow trends. This article adds the industry's perspective to the introductory article on trend following: regulation, position sizing, fees, crisis protection, dry spells and what studies say about returns after costs.

What CTAs are

The name comes from US futures law. According to the self-regulatory organisation NFA, a commodity trading advisor is a person or organisation that, for compensation, advises others on trading futures, directly or indirectly. CTAs that manage client accounts must register and become NFA members; the regulator, the CFTC, has delegated registration to the NFA. There are exemptions, for example for advisers who have advised no more than 15 persons in twelve months and do not hold themselves out to the public as a CTA (NFA 2026). Anyone who runs a fund that invests the money of many investors in futures is also a commodity pool operator.

Originally, CTAs mainly traded commodity futures; today financial futures dominate. In Europe, such strategies are often offered to retail investors as UCITS funds, and to professional investors also as alternative investment funds. The industry calls the asset class managed futures and its managers CTAs; industry indices are built from returns that funds report voluntarily to databases, with all the biases that entails (Bhardwaj et al. 2014).

Strategies: not just trend following

• Trend following: the largest part of the industry. Signals from past returns, moving averages or breakouts across many markets and time horizons; the article on trend following explains the basics. Levine and Pedersen (2016) show that time-series momentum and moving-average crossovers are equivalent in their general form. Many rules are variants of the same idea.
• Carry: long in markets with a high running yield, short in those with a low one, for example in currencies, bonds or commodities in backwardation. More in the article on the carry trade.
• Mean reversion and short-term models: bets on countermoves over days or hours, often as a complement to trend following.
• Systematic or discretionary: most CTAs trade according to fixed rules. Discretionary traders decide for themselves, often based on their view of the economy; these funds are close to the global macro strategy.

Who earns when trend followers earn? In the position data of US futures markets, Moskowitz et al. (2012) found that speculators profit from time-series momentum at the expense of hedgers, i.e. producers and consumers who want to offload price risk. Lempérière et al. (2014) document trends in four asset classes since 1800 with very high statistical significance; long-term trends have not weakened in recent times, whereas short-term ones have weakened markedly.

Position sizing: volatility target and risk budget

The direction comes from the signal, the size from the risk. Many CTAs steer towards a target volatility for the whole portfolio and allocate it to the markets as a risk budget. As a result, calm markets get a much larger notional value than volatile ones.

For the futures, the fund only has to post margin, usually a small fraction of the notional value. The rest sits in money market instruments and earns interest, which is an often overlooked part of the return. Kim et al. (2016) show how important the scaling is: the strong results of time-series momentum depend largely on scaling by volatility; without it, the strategy behaves much like buy and hold. The glossary describes the volatility target as a rule for an entire portfolio under target volatility.

Crisis alpha and convexity

Trend followers have a special reputation: they often make money precisely when equities and bonds fall for a long time. The industry calls this crisis alpha. Fung and Hsieh (2001) offered an explanation: the returns of trend followers resemble those of a lookback straddle, an option that pays out the difference between the highest and the lowest price over a period. Such options explain the returns better than linear models with market indices.

Dao et al. (2017) show that the outcome of trend following depends on the difference between variance measured over long and over short horizons: when moves extend in one direction over months, long-term variance is greater than short-term variance, and the trend follower wins. This gives rise to the positive convexity of industry returns, which is stronger than previously thought when measured correctly. According to Moskowitz et al. (2012), time-series momentum performed best precisely in extreme market phases, and Hurst et al. (2017) find positive results over more than a century in most of the major crises of a portfolio of equities and bonds.

The protection comes with conditions. It takes time for the signals to turn; a one-day crash after a long rally catches the trend follower positioned long. And it protects against prolonged moves, not against every decline.

Dry spells: the 2010s

After the large gains of 2008, a lot of money and attention flowed into managed futures, and the results of the following years were below the long-term average. Hutchinson and O’Brien (2014) examine this pattern with almost a century of data: in the years after financial crises, trend following earned on average less than half the return of crisis-free periods, and it was particularly weak in the four years after the start of a crisis. Futures markets then showed hardly any of the usual continuation of trends.

Baltas and Kosowski (2020) examine how strongly trend strategies depend on the correlations between markets, with a view to the weak years after the financial crisis: the more markets move together, the fewer independent trends there are. On top of that, interest rates were near zero, so the collateral hardly earned anything either. Dry spells lasting several years are nothing unusual for trend followers. Those who cannot sit them out typically exit shortly before the next good phase.

Costs: 2 and 20, replication, funds

The classic fee model is called 2 and 20: a 2% management fee a year and a 20% performance fee on gains above the high-water mark, the previous peak. If the performance fee is charged on the total return, it also applies to the interest on the collateral, which you would earn without a manager anyway.

Bhardwaj et al. (2014) measured how much this matters in practice for 1994 to 2012: before fees, CTAs earned 6.1% a year above US money market instruments; after fees, the excess return was not significantly different from zero. The managers kept the bulk of it. Even before fees, the authors found no alpha relative to simple, publicly known futures strategies. This gave rise to the idea of replicating the core of the industry cheaply. Braun et al. (2024) show that an index of trend-following funds can be replicated, but warn that a simple regression brings its own tracking risk, because the fee savings have to be weighed against the loss of manager skill.

Besides the fees, trading and rolling the futures also cost money. Baltas and Kosowski (2020) split these costs into roll costs and rebalancing costs and show how better volatility estimators and signals reduce turnover. For euro investors, funds in dollars add currency risk if the share class is not hedged.

Role in the portfolio: allocation or overlay

Trend following can be used in two ways. As an allocation (a carve-out), it replaces part of the portfolio. As an overlay, it is added on top via futures without tying up capital; you then only earn the return above the money market. The second variant is also called return stacking. The example uses assumptions, not historical returns.

Because trend following often gains in long downturns, it acts like a reserve during rebalancing: after a crisis, money is moved out of it into the assets that have fallen. The risk contribution shows how much of the overall risk a component carries. Whether you should use trend signals yourself to time your own portfolio is discussed in the article on market timing; the topic of momentum strategies is also related.

What the research shows

Long data series support the strategy: Moskowitz et al. (2012) found time-series momentum in all 58 futures examined, and Hurst et al. (2017) and Lempérière et al. (2014) over more than one and two centuries respectively. Fung and Hsieh (2001) and Dao et al. (2017) explain the convexity. But there are serious objections. In regressions for each asset, Huang et al. (2020) find little evidence that the return of the past twelve months predicts the next one; a strategy based only on historical means earns similar returns. Kim et al. (2016) attribute the success largely to scaling by volatility.

For investors, the gap between the strategy and the product also matters. Bhardwaj et al. (2014) show that fees consumed most of the CTAs' gross return, and Hutchinson and O’Brien (2014) that good crisis years are often followed by lean ones. The sober summary: trend following has a long, documented history as a diversifier in prolonged crises. How much of that reaches the investor after costs depends heavily on the product.

Advanced: correlations in the risk budget

The formula b = target / √N only holds if the markets are uncorrelated. With an average correlation ρ̄ between the markets' contributions, the portfolio's volatility rises:

Other levers are the volatility estimator (short windows react quickly but increase turnover), the mix of time horizons and the universe. Levine and Pedersen (2016) represent trend filters as weights on past returns; this makes it possible to compare which part of the past a rule emphasises. To evaluate a fund, a regression on simple trend factors helps, similar to style analysis: if no alpha remains after the factors are deducted, you are paying for beta.

Limitations and common mistakes

• Biased industry figures: CTA indices are based on voluntary reporting; poor funds stop reporting. Bhardwaj et al. (2014) therefore consider the reported results to be too good.
• Not every CTA follows trends: carry, short-term and discretionary strategies behave differently in crises.
• Delayed protection: fast crashes after long rallies catch trend followers positioned long; protection only emerges once the downtrend persists.
• Underestimated costs: management and performance fees, trading and roll costs, currency effects.
• Track record too short: many funds for retail investors have only existed for a few years. Simulated back-calculations invite backtest overfitting.
• Staying the course: dry spells lasting several years are normal. Those who sell after weak years often miss exactly the benefit in the next crisis.

This article is not investment advice and does not recommend any individual funds.

How Portfolio Lab calculates

The Tactical tool replicates trend and momentum rules long-only on a monthly basis: moving average, dual momentum, relative strength and adaptive allocation. The “Target volatility” model applies CTA-style position sizing to an entire portfolio: it scales the risk portfolio so that the realised volatility of the last few months (default 6, selectable from 3 to 36) matches the target (default 10%, selectable from 2 to 30%). The investment ratio is capped at 100% by default and can be raised to 200%; anything above 100% appears as a negative cash share. Signals are based only on prices up to the month-end. There are no short positions and no futures, so the Tactical tool cannot show gains from falling markets the way a CTA can.

For the product view, Asset classes offers the class “Trend following (managed futures)” (@TRENDFOLGE), chained from two US managed futures funds from 2010 onwards and converted into euros, i.e. without currency hedging; the fund costs are included in the series. In the Backtest, you can use it to test an allocation; with the leverage per security, you can also approximate an overlay. Details are in the Tactical methodology.

This article explains a method; it is not investment or tax advice. Worked examples are simplified and rounded. All terms in the glossary.