A track record is the most heavily used piece of evidence in manager selection, and among the easiest to misread. The same monthly series supports sharply different conclusions depending on how it was built, how long it runs and what it is set against. This guide sets out the checks we apply before a CTA's numbers go into an investment committee paper.
Start with the basis of the numbers
Before calculating any ratio, establish what each month in the series represents. Three categories recur:
- Actual performance. Results of real client accounts or of a fund traded under the programme, net of fees and expenses.
- Proforma performance. Actual results restated for a difference in terms, such as the current fee schedule or another leverage level.
- Hypothetical performance. Results of applying a model to historical data before it traded client money, or of trades that were never executed.
These distinctions carry regulatory weight. CFTC Regulation 4.35 governs the performance a registered CTA presents in its disclosure document, and NFA Compliance Rule 2-29 restricts the use of hypothetical results in promotional material and requires a prescribed cautionary statement alongside them. [1] [2] Simulated periods benefit from hindsight in model design and carry no slippage, execution cost or operational error. Treat them as a description of the model's logic, not as evidence of what an investor would have experienced.
Composites and account size
Where a CTA trades many separate accounts, the reported record is usually a composite. Ask how accounts enter and leave it, whether smaller accounts trade a reduced set of markets, and how partially funded accounts are treated. NFA rules on notional funds require returns for accounts funded below their nominal size to be computed on a consistent basis, because returns measured on cash alone would overstate both gains and losses. [3] The mechanics are covered in Understanding CTA Managed Accounts.
How long is long enough?
Short records carry wide statistical uncertainty. Andrew Lo's 2002 paper on the statistics of Sharpe ratios gives an approximate standard error for the estimate when returns are independent and identically distributed. [4] Applied to an observed annualised Sharpe ratio of 0.8, the approximate 95% interval runs from about −0.50 to 2.10 after three years of data, and narrows to about 0.09 to 1.51 after ten. Serial correlation in monthly returns changes these intervals, and Lo sets out the adjustment.
Length matters for a second reason: exposure to distinct market environments. A record spanning the 2008 financial crisis, the long period of near-zero policy rates that followed, the March 2020 liquidity shock and the 2022 tightening cycle has met conditions that a five-year record has not. For newer programmes, process evidence carries more of the weight: research method, risk controls, and the prior records of the people running the money.
Risk-adjusted ratios and their limits
The Sharpe ratio divides excess return by volatility. The Sortino ratio replaces total volatility with downside deviation. The Calmar ratio, introduced by Terry Young in 1991, divides annualised return by maximum drawdown. [5] Each answers a narrow question, and each depends on choices that should be disclosed:
- The risk-free rate deducted, which matters more when short-term rates are high and interest on collateral forms part of the return.
- The window. Full-history, three-year and five-year figures often rank the same programmes differently.
- Return frequency. Monthly data smooths intra-month volatility and drawdowns.
Volatility-based ratios treat upside and downside dispersion alike. For programmes with asymmetric return distributions, read them next to skewness and the measures in Drawdown and Recovery Metrics Explained.
Database effects: survivorship and backfill
Peer comparisons inherit the biases of the database behind them. Managers report to commercial databases voluntarily. Programmes that close after poor results tend to stop reporting, which lifts the average of those that remain; managers joining a database with an existing history may contribute their stronger early years. Fung and Hsieh documented both effects in hedge fund and commodity fund data in 2000. [6] When a programme is ranked against peers, ask whether the peer set retains defunct programmes and how backfilled history is handled.
Fees, assets and capacity
Confirm that returns are net of management and incentive fees, and that the fee schedule shown matches the terms on offer. Incentive fees are normally subject to a high-water mark, so an investor who enters after a loss pays different effective fees from one who enters at a peak. Read assets under management next to returns. Rapid asset growth may push a programme towards the most liquid contracts, change its market mix and alter its behaviour relative to its early years.
Behaviour in stress periods
Full-period averages conceal the months that matter most to a portfolio. Examine the monthly return table for named episodes, such as the fourth quarter of 2008, the first quarter of 2020 and the whole of 2022, and measure correlation to equities in the worst equity months as well as across all months. Our study of crisis alpha shows how these characteristics cluster across the Gnosis manager universe.
A working checklist
- Is every month labelled actual, proforma or hypothetical?
- Do the figures reconcile to the CFTC disclosure document and, for pools, to audited financial statements?
- Are returns net of all fees, on the fee basis offered?
- How many years of actual trading exist, and which market regimes do they cover?
- How do Sharpe, Sortino and Calmar ratios move across one, three and five years and full history?
- What were the depth, duration and recovery time of the largest drawdowns?
- How have assets under management changed, and has the market mix shifted with them?
Allocator IQ standardises more than 20 metrics per programme across its manager universe and marks hypothetical periods, so these comparisons start from a consistent base. The platform page shows the workflow.
Frequently asked questions
What is the difference between actual and hypothetical CTA performance?
Actual performance comes from real client accounts or funds traded under the programme, net of fees. Hypothetical performance applies a model to historical data, or reflects trades that were never executed, and benefits from hindsight. NFA rules require hypothetical results in promotional material to carry a prescribed cautionary statement.
How many years of track record should a CTA have?
No regulatory minimum exists, and many allocators set their own threshold. Statistically, a ratio estimated over three years carries a wide margin of error, and a longer record is more likely to cover distinct regimes such as 2008, 2020 and 2022.
What is a VAMI?
VAMI, the Value Added Monthly Index, shows the growth of a notional 1,000 investment by compounding a programme's monthly net returns. It is the standard growth-of-capital chart in managed-futures reporting.
Should a CTA track record be compared with an index?
Yes, with care. A CTA index or peer group gives context, but survivorship and backfill biases affect peer averages, and an index may differ from the programme in market mix and volatility.
For institutional investors. This article provides general information, not investment, tax or legal advice or a recommendation of any manager, strategy or structure. Futures trading involves substantial risk of loss and is not suitable for all investors. Past performance is not necessarily indicative of future results.
Sources
- CFTC Regulations, 17 CFR 4.35. Performance disclosures of commodity trading advisors.
- NFA Compliance Rule 2-29. Communications with the public and promotional material, including hypothetical results.
- NFA Interpretive Notice 9054 and Compliance Rule 2-34. Notional funds and rate-of-return computation for partially funded accounts.
- Lo, A. W. (2002). The Statistics of Sharpe Ratios. Financial Analysts Journal, 58(4).
- Young, T. W. (1991). Calmar Ratio: A Smoother Tool. Futures magazine.
- Fung, W. and Hsieh, D. A. (2000). Performance Characteristics of Hedge Funds and Commodity Funds: Natural vs. Spurious Biases. Journal of Financial and Quantitative Analysis, 35(3).