Robert Carver's Systematic Trading: the ideas to test on futures

Contents

1. What are the main ideas in Robert Carver's Systematic Trading?
2. How is Carver's volatility target calculated?
3. How does Carver decide position size?
4. Why diversify across rules and instruments?
5. Why keep rules simple, and what does a simple rule look like in a test?
6. Is Systematic Trading suitable for retail traders?
7. How does Carver differ from Simons and Chan?
8. Frequently asked questions
9. References

What are the main ideas in Robert Carver's Systematic Trading?

Robert Carver's Systematic Trading is a framework for running a trading account by written rules, not a collection of entry setups. Its main ideas are: size every position from a volatility target, combine many simple rules across many instruments, trade slowly enough that costs stay small, and let no decision be made by feel.

The book treats a strategy as a chain. A trading rule produces a forecast. A forecast is a number that says how strongly the rule wants to be long or short right now, not just a yes or no. The forecast is scaled to a standard range so that different rules can be compared. The volatility target turns the account into a cash amount of risk. The instrument's own volatility turns that cash risk into a number of contracts. Finally, weights combine the positions of all rules and all instruments into one portfolio.

A good way to picture the chain is a kitchen with a recipe card for every dish. The entry rule is one line on the card. Most of the card is about portions, timing and how the dishes fit together on one plate. That is where Carver spends most of his pages, and it is why the ideas below are about size, mix and speed more than about entry rules. The same three settings are what a futures backtest must get right before its result means anything.

How is Carver's volatility target calculated?

A volatility target is the yearly risk you choose for the whole account, written as a percent of capital, and the position in one instrument is that cash risk divided by the instrument's own yearly volatility in cash per contract. A quiet market therefore gets more contracts and a wild market fewer, so each position carries about the same risk.

Volatility here means the typical size of yearly price swings, measured as a standard deviation. Think of it as the width of the road the price usually drives on. Carver measures it from recent returns, so it changes as the market changes. The one formula to keep:

contracts = (account × volatility target) ÷ (yearly volatility of one contract in cash)

Here account is the capital in the account, volatility target is the yearly risk you chose as a fraction of that capital, and yearly volatility of one contract in cash is how much one contract typically moves in a year, in the same currency. The top of the fraction is the cash risk you want; the bottom is the cash risk one contract delivers. To check the contract count for a single trade from a stop distance instead, the free position size calculator does the division for you.

The target itself is a choice, not a calculation. Carver links it to how confident you are in the strategy and to the Kelly criterion, which gives the risk level that grows capital fastest when the edge is known exactly (Kelly, 1956), and he argues for trading well below that level because the edge is never known exactly. Pick a target, write it down, and keep it the same across every instrument and every test.

How does Carver decide position size?

Position size follows from the volatility target and the scaled forecast, never from a fixed contract count. Position size is the number of contracts held in one instrument at one moment. Carver scales every forecast so that its average strength is 10 and its largest allowed value is 20. A forecast of 20 holds twice the contracts of a forecast of 10, and a forecast near zero holds almost nothing. The volatility target sets the contracts at an average forecast; the forecast then scales that number up or down.

A fixed contract count and a volatility-scaled count can take exactly the same trades and still carry very different risk. In a calm stretch the fixed count is too small; in a volatile stretch it is too large, and that is usually when the deep drawdowns arrive. The figure shows the pattern on an illustrative timeline.

To show this on real trades, AlphaPrime (Valynox) reran one set of ES daily-bar trades under nine money-management methods with $100,000 of capital and at most 10 contracts, using each method's default settings in the report. The entries and exits are the same in every row. Only the contract count changes, so every difference in profit and drawdown comes from sizing alone. The trades come from the candidate described in the section on simple rules: one kept out of 1,024 candidates for its return against its drawdown, which flatters every row below, so none of the figures is what the rule earns.

Three things to read in the box. First, the fixed-dollar-risk and volatility-target rows land in the same place, because both divide a fixed cash risk by a measure of how much the instrument moves. Second, the fixed-fractional-risk row shows a maximum drawdown of $318,270 against $100,000 of starting capital, which means the account would have been emptied before the profit arrived; a high average contract count bought that profit with a drawdown no account survives. Third, the Kelly row held about one contract on average and kept the drawdown at $57,590, the same figure as the one-contract run. Sizing changed the drawdown by a factor of more than five without changing a single trade.

Why diversify across rules and instruments?

Combining rules and markets that do not move together lowers portfolio volatility more than improving any single rule does, which is why Carver spends more pages on correlation and weights than on entry signals. Correlation is a number from minus one to plus one that says how closely two return streams move together. Two rules that each lose in the same weeks give no relief; two rules that lose in different weeks give a combined curve with shallower dips. This is the idea Markowitz (1952) formalized for portfolios of assets, and Carver applies it to portfolios of rules as well as instruments.

In practice the framework does this in two layers. Forecast weights say how much each rule counts inside one instrument. Instrument weights say how much each instrument counts in the account. Carver's examples lean on trend-following and carry rules, which tend to have low correlation with each other, across futures in equities, bonds, currencies and commodities. He also argues for simple weighting schemes, such as splitting weight evenly among groups of similar things, because precise weights estimated from history are mostly noise.

For a tester, the practical consequence is that the number to read is the combined drawdown of the portfolio, not the drawdown of the best single strategy. A rule that looks ordinary on its own can be the one that smooths the account because it loses at different times from the others. How to run strategies together and read their combined curve is covered in portfolio backtest: the combined drawdown is the one you live through.

Why keep rules simple, and what does a simple rule look like in a test?

A rule with one or two parameters has fewer ways to fit noise and gives a search fewer variants to try, so its best result carries less selection bonus. Carver's own rules are deliberately plain: a difference of two moving averages for trend, a measure of roll yield for carry. He argues that a rule you can explain in one sentence is one you can keep trading through a losing year, and that the gain from a cleverer rule is usually smaller than the gain from adding another uncorrelated one. Bailey et al. (2014) give the statistical side: the more variants a search tries on one history, the more the best of them is inflated by luck.

To show what a single-indicator rule looks like over several years, the box below reports a run on ES daily bars from 2019-01-02 to 2025-06-30, long and short, one contract, with $20 commission and $25 slippage per trade. The rule is a stand-in, not one from Carver's book: a single Average True Range (ATR) rule serves as the only entry and exit, with no stop, no target and no filter. It is the one candidate the search kept out of the 1,024 candidates it tried, chosen for its return against its drawdown. Picking the kept candidate after the fact flatters its figures, so they are not what the rule earns and not what a reader can expect.

What the box shows is the shape of a slow, always-in rule: 165 trades in six and a half years, in the market 82.9 percent of the time, with a maximum drawdown about half the size of the net profit. That proportion is typical of plain trend-style rules and is the reason Carver pairs them with other rules rather than trading one alone. When the search settings limit a rule to one or two parameters, the walk-forward and threshold checks in strategy validation are applied to far fewer variants, and the gap between the kept candidate's in-sample and out-of-sample figures is the number to watch.

Is Systematic Trading suitable for retail traders?

Yes for a futures trader with enough capital to hold several instruments at the target risk; the main gap between Carver's framework and a retail account is that contracts come in whole units. The framework asks for a low volatility target spread across many markets. Each market then gets a small slice of cash risk, and if one contract already carries more risk than that slice, the account cannot hold that market at all.

Two things narrow the gap. Micro contracts, such as the Micro E-mini equity index futures at one tenth the size of the E-mini, let a smaller account hold more markets at the target risk (CME Group FAQ). And Carver's later book, Leveraged Trading, starts from one rule on one instrument and adds rules and markets only as capital allows, which is the same framework built from the bottom up. A retail account that holds a single stock index future is close to a leveraged buy-and-hold position, and the yearly swings that position carries are the ones described in average stock market return: what it hides for systematic traders.

What a retail trader can copy without any extra capital: a written volatility target, a fixed cash risk per instrument divided by measured volatility, and a trading speed slow enough that the round-trip cost is a small share of the average trade. These three settings change a backtest more than any entry rule does.

How does Carver differ from Simons and Chan?

Carver trades slow, diversified trend and carry rules inside a risk framework one person can run; Jim Simons's Renaissance combined very many short-horizon signals with a large team and data budget; Ernest Chan writes his own code for mean-reversion and statistical-arbitrage strategies, usually one strategy at a time. The three differ in speed, breadth and what the individual must build.

Dimension	Carver	Simons (Renaissance)	Chan
Trading speed	Days to weeks	Intraday to days	Intraday to days
Number of rules	Dozens of simple rules	Very many signals	A few, tested one by one
Instruments	Many futures markets	Many asset classes	Stocks, currencies, futures pairs
Risk framework	Volatility target and weights	Firm-level, not published	Kelly-based sizing
Who does the work	One person with a spreadsheet or script	Team of researchers and engineers	One person who codes
What retail can copy	The whole framework at smaller scale	The research discipline, not the scale	The strategy types and the testing habits

The difference that matters most for a reader is the size of the operation. What a single trader can take from Renaissance is covered in Jim Simons: what a retail trader can actually copy from Renaissance, and the wider gap between a quant firm and a home office in quantitative trading firms: what they do and what retail can copy. Chan's approach is closest to a solo trader who can program; his testing habits are in Ernest Chan's Quantitative Trading: lessons for backtesting. Carver's book is the one that assumes no team, no large data budget and no custom code, which is why its ideas transfer most directly to a retail futures account.

A first thing to do with the book: take one strategy you already have, hold its entries and exits fixed, and rerun it under a fixed contract count and under a volatility-scaled count, as the second box in this article does. Then compare the two maximum drawdowns. If they differ by more than you expected, the sizing rule, not the entry rule, is the first thing to write down.

Frequently asked questions

What is a forecast in Carver's framework?

A forecast is a number that says how strongly a trading rule wants to be long or short at the moment, not a simple buy or sell. Carver scales every forecast so its average strength is 10 and its largest allowed value is 20. The position in an instrument is proportional to the forecast, so a strong forecast holds more contracts than a weak one at the same volatility target.

Does volatility targeting raise returns?

Volatility targeting aims to hold the risk of each position steady, not to raise returns. It gives more contracts in quiet markets and fewer in volatile markets, so the account's swings stay close to the chosen target. In the ES rerun in this article, the volatility-target row changed the maximum drawdown while taking exactly the same trades; the entries and exits did not change at all.

How many instruments does Carver's framework need?

The framework works with one instrument and gains most of its smoothing from the first few uncorrelated additions. The practical limit is capital: each instrument needs enough cash risk for at least one whole contract at the chosen volatility target. Micro contracts lower that threshold, and Carver's later book Leveraged Trading describes starting with one rule on one market and adding more as capital allows.

What is the difference between Systematic Trading and Leveraged Trading?

Systematic Trading presents the full framework: forecasts, volatility targeting, rule and instrument weights, and the written process behind them. Leveraged Trading applies the same ideas to smaller accounts and to leveraged products, beginning with a single rule on a single instrument and adding complexity step by step. Readers with a small futures account often find the second book an easier starting point.

References

• 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.
• Markowitz, H. (1952). Portfolio Selection. The Journal of Finance, 7(1), 77–91.
• Kelly, J. L. (1956). A New Interpretation of Information Rate. Bell System Technical Journal, 35(4), 917–926.
• CME Group. FAQ: Micro E-mini Equity Index Futures. https://www.cmegroup.com/articles/faqs/frequently-asked-questions-micro-e-mini-equity-index-futures.html