CTA strategy explained: how trend-following futures funds work
A CTA strategy is a written rule set a commodity trading advisor runs on futures for clients. Most CTAs follow trends across dozens of unrelated markets, long and short, size each position by its volatility, and accept long drawdowns in return for a few strong trending years. A retail trader can test a simple version with a few liquid futures, one trend rule, realistic costs and an out-of-sample period.
By Soren Huang · Published · updated · Editorial policy
Contents
What is a CTA strategy?
A CTA strategy is a rule-based way of trading futures that a commodity trading advisor runs for clients, and most CTAs follow trends: they buy markets that are rising, sell short markets that are falling, and spread the same rule across dozens of unrelated futures so that a few strong trends pay for many small losses.
A CTA (commodity trading advisor) is a person or firm that is paid to trade futures, or to advise on futures trading, on behalf of other people. In the United States the name is a regulatory category: a manager who does this for a fee usually has to register with the futures regulator, the CFTC, and give clients a disclosure document that describes the trading program and its past record. The client's money normally stays in the client's own futures account at a broker. An advisory account is an account that belongs to the client but that the CTA has written permission to trade. Managed futures is the name of the investment category that these programs belong to.
The word "commodity" in the name is historical. A modern CTA trades equity index futures, government bond futures, currency futures, energy, metals and grains, and the physical commodities are often the smallest part of the basket. What ties the basket together is one rule set, not one market. In that sense a CTA strategy is a large-scale form of algo trading: rules written in advance and run without discretion, applied to many markets at once.
Trend following is a strategy that enters a market in the direction of its recent move and holds until that move reverses, with no forecast of where the price will end. Most CTAs run some form of it. Unlike many stock trading algorithms, which hold long positions and pick which stocks to own, a CTA program is as comfortable short as long, because a futures contract is sold short as easily as it is bought.
How does a trend following CTA actually trade?
A trend following CTA buys after a market has already started to rise, sells short after it has already started to fall, holds for weeks or months, and runs this same rule on every market in its basket at the same time. The entry is late on purpose and the exit gives back part of the profit on purpose; both are the price of never having to predict a turn.
The life of one trade has four steps. First, a signal: the price closes above its highest close of the last N days, or a short moving average crosses above a long one. Second, the entry: the position is opened on the next bar, not on the bar that produced the signal, because the signal is only known at that bar's close. Third, the hold: the position stays open as long as price stays on the right side of a trailing line, usually a moving average or a trailing stop placed a few units of volatility below the recent high. Fourth, the exit: when price closes below that line, the position is closed and, in many programs, reversed. The inside bar breakout strategy on futures is a short-term relative of the first step; a CTA uses the same idea with a much longer lookback.
| Entry rule | Signal | Typical lookback | Strength | Weakness |
|---|---|---|---|---|
| Breakout | Close above the highest close of the last N days (long) or below the lowest (short) | 20 to 100 days | Never misses a large move; simple to code | Many false breakouts in a sideways market |
| Moving-average cross | Short average crosses the long average | Pairs such as 50 and 200 days | Smoother; fewer whipsaws | Enters later and exits later than a breakout |
Holding for weeks or months is the opposite of day trading, which closes every position before the session ends. A CTA accepts overnight gaps and weekend risk because the trends it wants to catch last far longer than a session. Moskowitz, Ooi and Pedersen (2012) tested a rule of exactly this kind, long when the past twelve months' return was positive and short when it was negative, across dozens of liquid futures and forward markets, and found the pattern present in nearly every one of them.
The basket is not decoration. In any given year most markets move sideways and the rule loses a little in each of them. The two or three markets that trend strongly have to pay for the rest. This is why a CTA would rather hold one contract in each of fifty unrelated markets than fifty contracts in one.
What returns do CTA funds make?
CTA results are lumpy by design: long flat or slowly losing stretches broken by a few strong trending years, with low correlation to stock indexes, so the shape of the return stream matters more than any single average figure. Two funds with the same average return can be very different to hold, and a trend follower is almost always the harder one.
The reason is in the trade structure from the previous section. Most trades are small losses taken when a breakout fails. The profit arrives in a minority of trades that run for months. Those trades cluster in years when many markets trend at once, which historically has often meant years of financial stress, when stocks fall, bonds rally and currencies move far. Moskowitz, Ooi and Pedersen (2012) found that the returns of their trend rule were strongest when stock markets made their most extreme moves, in either direction, and weakest when stocks were calm. A CTA record therefore looks nothing like a savings account: it looks like a staircase with long landings.
This has a practical consequence for anyone comparing funds or backtests. Ask for the full equity curve by month, not a summary number. Look at how long the longest flat stretch was, how many years produced most of the gain, and whether the gains came in the years when a stock portfolio would have needed them. A CTA that made all its money in one year and lost a little in each of the other nine may still be a useful part of a portfolio; a smooth-looking average hides that.
How do CTAs size positions and control risk?
CTAs scale each position by the market's recent volatility so that every market risks a similar small fraction of capital per day, which is why a quiet bond future gets more contracts than a volatile index future. Without this step the basket would be dominated by whichever market happened to be moving most, and the point of holding many markets would be lost.
Volatility-based sizing is a rule that sets the number of contracts from how much the market typically moves in a day, so that the expected daily swing of each position in dollars is about the same. The Turtle traders described by Covel (2007) used a version of it in the 1980s, measuring the daily range with an average true range and buying fewer contracts when the range was wide. The arithmetic needs one formula:
contracts = target daily risk ÷ typical daily move per contract
Here target daily risk is the dollar amount the manager is willing to see one position swing on a normal day, often a fraction of one percent of the account; typical daily move per contract is the recent average daily range of one contract, converted to dollars with the contract's point value. Divide one by the other and round down. The same arithmetic, with a stop distance in place of the daily move, is what the free position size calculator does for a single trade.
Hypothetical example
A manager sets a target daily risk of $2,000 per position. A quiet bond future moves about $500 per contract on a normal day, so it gets $2,000 ÷ $500 = 4 contracts. A volatile index future moves about $2,000 per contract per day, so it gets $2,000 ÷ $2,000 = 1 contract. Both positions now swing by about $2,000 on an ordinary day, and a trend in either one contributes about equally to the account.
Two more controls sit above the single position. The first is a cap on total risk across the basket, so that when many markets trend together, which also means they are correlated, the manager cuts size rather than letting the account's daily swing multiply. The second is a re-sizing rule: when a market's volatility doubles, the position is trimmed, usually at the next signal rather than every day, so that the dollar risk stays near target without constant trading costs.
How big are CTA drawdowns?
Trend following has a low win rate and long drawdowns measured in months or years, so a trader must plan for drawdown length as much as depth and expect the live drawdown to exceed the backtest's. The depth is the number everyone looks at; the length is the number that makes people quit.
A drawdown is the fall in account value from its previous highest point to the lowest point that follows, before a new high is made. It has a depth, the size of that fall, and a length, the time from the old peak to the new one. For a trend follower the length is long because the method only makes money when markets trend, and it cannot force them to. A sideways year produces a slow bleed of small losses and no way to recover until the next trend arrives. Because exits happen only after a reversal, fewer than half of the trades are typically winners, and the profit comes from a few large ones; a long run of losing trades is therefore normal, not a sign that the rule is broken.
| Depth | Length | |
|---|---|---|
| What it measures | Peak-to-trough fall in account value | Time from the old peak to a new peak |
| Unit | Dollars or percent of the account | Months, sometimes years |
| What it threatens | Margin and account survival | Patience and the decision to stop trading |
| What to check in your own test | Maximum drawdown against the capital you would commit | Longest flat stretch against the longest you have ever held a strategy unchanged |
Expect the live figure to be worse than the backtest on both counts. The backtest shows one path through history; live trading shows a new one, and the backtest you chose to trade was selected partly because its drawdown looked acceptable. Magdon-Ismail and Atiya (2004) show that the expected maximum drawdown of an equity curve grows with the length of the record, which is why a long-running CTA will at some point show a deeper drawdown than any in its first years. The free drawdown calculator reads depth and recovery time from any equity series you paste in, so you can measure both for a backtest or a fund record before deciding whether you could hold it.
Can a retail trader test a simple version?
Yes: pick a few liquid futures, one trend rule, volatility-based sizing, realistic commission and slippage and an out-of-sample period, then judge the result against the drawdown you can hold. The test does not need fifty markets to teach you how trend following feels; three to five markets and ten years of daily bars show the flat stretches and the lumpy gains clearly enough.
- Choose three to five liquid futures from different groups, for example one index future, one bond future, one energy and one metal, so that they do not all trend at once.
- Pick one trend rule and write its parameters down before you run anything: a breakout of N days with a trailing exit, or a moving-average cross, not both.
- Set volatility-based sizing as in the previous section, with a small target daily risk, so that no single market dominates.
- Set commission and slippage per contract at the level your broker actually charges, plus at least one tick of slippage per side.
- Split the history into an in-sample period for choosing the parameters and an out-of-sample period that you run once, unchanged.
- Judge drawdown length and depth against the capital and the patience you really have, before you look at the profit.
To show what the output of such a test looks like, we ran one on a single market. The run used NQ daily bars from 2019-01-02 to 2025-06-30, long and short, one contract, with commission and slippage charged on every trade. Its entry and exit came from a Choppiness Index rule with no stop, target or filter. That is a stand-in built from a standard indicator, not a CTA breakout rule and not a basket, so it shows the shape of a multi-year futures test rather than the subject itself. Make the costs part of every evaluation rather than a deduction at the end; the backtesting in AlphaPrime (Valynox) charged the commission and slippage inside each candidate and filled on the next bar, which is the setting this run used.
Source: AlphaPrime product engine (CPU), run 2026-10-31. NQ daily bars, 2019-01-02 to 2025-06-30, long and short, 1 contract, $20 commission and $10 slippage per trade.
| Number | Result |
|---|---|
| Net profit (closed trades) | $221,050 |
| Max drawdown | $75,190 |
| Trades | 45 |
| Win rate | 64.44% |
| Profit factor | 2.354 |
| Average trade | $4,912.22 |
| Time in market | 56.75% |
Read these numbers the way the earlier sections suggest. This is the one candidate the search kept out of 1,024 it tried, chosen for its return against its drawdown, and picking the kept one after the fact flatters it: the best of a thousand tries always looks better than the rule that produced it, which is the selection effect Bailey, Borwein, López de Prado and Zhu (2014) describe. The rule did not earn this money and a reader should not expect it. The win rate of 64.44% is higher than classic trend following usually shows, for two reasons: a Choppiness Index rule enters and exits on a different signal than a breakout, and 45 trades is a small sample, so a few trades either way would move that figure a long way. The max drawdown of $75,190 against a net profit of $221,050 over about six and a half years is the kind of ratio to compare with your own capital: a trader with a small account would have been forced out well before the end.
Your next step is to repeat the six steps above on one liquid future you already follow, with a breakout rule instead of the stand-in, and to read the longest drawdown in months from the report before you read the profit. If that length is longer than any period you have held a strategy unchanged, the rule is not the problem; the plan for sitting through it is.
Frequently asked questions
What does CTA mean in trading?
CTA stands for commodity trading advisor. It is a person or firm paid to trade futures, or advise on futures trading, for other people, and in the United States it is a regulatory category that usually requires registration with the CFTC. Despite the word commodity, most CTAs today trade mainly financial futures such as equity indexes, bonds and currencies.
Do CTAs only trade commodities?
No. The name dates from a time when futures exchanges listed mostly agricultural and metal contracts. A modern CTA basket is usually dominated by equity index, bond and currency futures, with energy, metals and grains making up the rest. The method is the same in every market: one trend rule, long or short, sized by volatility.
Why do CTA funds often do well when stocks fall?
A trend follower holds whatever has been moving, in either direction, and large stock-market falls tend to come with strong moves in bonds, currencies and commodities as well. Moskowitz, Ooi and Pedersen (2012) found that a simple trend rule across many futures markets had its strongest returns in the periods when stock markets moved most, up or down. This is a pattern in past data; whether it appears again in any given fall is not known in advance.
What win rate does a trend following CTA have?
Usually fewer than half of the trades are winners. The method exits only after a trend has reversed, so every failed breakout is a small loss and the gains come from a minority of trades that run for months. A long run of losing trades is a normal part of the method, which is why drawdown length matters as much as depth.
Can I run a CTA-style strategy with a small account?
You can test one with a small account, but running it live needs enough capital to hold several markets at one contract each and to sit through the drawdowns the test shows. Micro contracts make the first part easier. The second part depends on the longest drawdown in your test, in months and in dollars, measured against what you can actually afford to lose and wait for.
References
- Moskowitz, T. J., Ooi, Y. H., & Pedersen, L. H. (2012). Time Series Momentum. Journal of Financial Economics, 104(2), 228–250.
- Covel, M. W. (2007). The Complete TurtleTrader: The Legend, the Lessons, the Results. HarperBusiness.
- Magdon-Ismail, M., & Atiya, A. F. (2004). Maximum Drawdown. Risk, 17(10), 99–102.
- Bailey, D. H., Borwein, J., López de Prado, M., & Zhu, Q. J. (2014). Pseudo-Mathematics and Financial Charlatanism: The Effects of Backtest Overfitting on Out-of-Sample Performance. Notices of the American Mathematical Society, 61(5), 458–471.