Choosing an investment strategy
How to evaluate an investment strategy's track record
A return number needs context: what was measured, over which dates, against which benchmark, and under what execution assumptions? Start with those questions before treating a strategy's record as a reason to follow it.
Published by Allocation Agents · Moore Tech, LLC. We operate the subscription described in this guide.
Identify what kind of record you are reading
A backtest applies rules to historical data. A forward simulation records decisions as time passes and models their execution. A real-account record reflects actual execution in an account. These sources of evidence have different limitations and should be labeled clearly.
Allocation Agents publishes simulated portfolios and hypothetical results, not customer-account performance. Its methodology and permanent decision records help explain what was modeled. A forward record avoids some hindsight problems, but it still depends on execution and cost assumptions.
Match the dates and understand record length
Check when the displayed strategy version began, the most recent valuation time, and the number of observations. A few weeks of performance are evidence about those weeks, not about an entire market cycle.
Do not compare a strategy's since-start return with a benchmark's year-to-date return. Use the same period. Be cautious about annualizing very short records: extending a recent pace mathematically is not a forecast.
Version changes and resets matter. Ask whether earlier records are preserved, whether results are linked across versions, and whether a change in mandate makes older performance less relevant to the current strategy.
Read benchmark-relative returns correctly
A benchmark gives context for what the market did during the same period. Check whether both sides include dividends and whether the comparison uses consistent valuation dates. Also consider whether the benchmark is relevant to the strategy's mandate.
For illustration only, a strategy returning 8% while its benchmark returns 10% trails by 2 percentage points. A strategy losing 3% while the benchmark loses 8% leads by 5 percentage points, but it still lost value. Relative performance and absolute gains are different questions.
Allocation Agents displays Return vs SPY in percentage points and uses SPY total-return comparisons. A defensive or cash-heavy approach may behave differently from a fully invested equity benchmark; read the mandate alongside the comparison.
Inspect drawdown, concentration, and losing decisions
Maximum drawdown describes the largest measured peak-to-trough decline within the observed record. It is not a limit on future losses. A short record may simply not have encountered the conditions that would stress the strategy.
Look at the largest positions, cash, sector exposure, and frequency of changes. Two portfolios with similar returns can have very different concentration and turnover. Inspect at least one losing or reduced position and read the explanation available at the time.
The profile's Changes view lets you examine recorded decisions rather than relying only on a summary number. Respect the dates and access labels: public delayed evidence is not a statement of today's holdings.
Check the assumptions between decision and result
Read how the model handles decision time, fill time, prices, fees, spreads, slippage, and dividends. A target allocation is not proof an order filled. Corrections should identify what changed and which records were affected.
If you act on a published strategy, your timing, prices, taxes, costs, and existing positions can differ. A simulated record therefore cannot establish identical subscriber results. Use it as one piece of evidence about the strategy, together with the mandate and explanations.
Finish with a practical question: can you trace a portfolio change from its recorded reason through modeled execution to its effect on the record? If the information is unavailable, treat that as a limit on what you can conclude.