Presenting live results separately from backtests

A manager may present backtests, demo-account outcomes and live results to explain a method. This supports review only when each section has a clear boundary. Joining curves without labels can make live experience appear longer or more reliable than it is. Chart styling should not conceal the type of evidence.

Define the three result types separately

A backtest applies model rules and assumptions to historical data. A demo account describes activity without real investment capital and may have different execution behavior. A live record comes from a real account’s recorded activity, but still requires source evidence, costs and observation limits. None automatically substitutes for another.

A hypothetical record-length example

Suppose a manager has 18 months of backtesting and 3 months of live trading. Displaying them together can help comparison, but calling the combination “21 months of live results” would be misleading. Show the boundary date, result type and any changes in method or capital on or beside the chart. This example describes no named manager.

What should a backtest disclose?

• Data source, time range, quality limitations and missing intervals.
• Method rules and when parameters were selected or changed, to the extent needed to understand the test.
• Assumptions for spread, commission, slippage, holding costs and executable size.
• The period used for tuning and any separate evaluation period.
• Excluded conditions, such as market interruptions or real-account restrictions.

Out-of-sample testing can provide additional information, not a guarantee of future results. If a supposedly separate period has repeatedly informed adjustments, disclose that rather than presenting it as completely untouched. The number of attempts and how a result was selected affect the interpretation.

What must remain identifiable in a live record?

Retain account identity and scope, start date, cash flows, costs, open positions and significant method changes in reconcilable records. Do not silently combine separate accounts into the history of a single account. Explain observation restrictions where the reviewer has limited access; polished presentation cannot replace reconciliation.

Comparison needs a defined basis

Specify currency, dates, exposure size and the return definition. Differences between simulated and live outcomes may arise from costs, delay, sizing or source data and do not alone establish one certain cause. Explanations should follow evidence rather than removing unfavorable observations without notice.

A revised model is not a rewritten live history

Record when and how a method changed after results were observed. A fresh backtest of the revised model is not that model’s past live record. Retaining versions helps a reviewer understand which information and rules were actually available to the manager at each point.

Use the account-manager cooperation route to explain record scope and reviewable evidence. In account information, do not present simulations as actual results. Submitting records or extending a chart neither purchases trust nor promises acceptance.

This is guidance for transparent presentation, not performance certification. Both backtests and live records have limitations. Separating them supports better judgment without guaranteeing future profits.