The starting brief
“Buy when the fast moving average crosses above the slow one. Sell on the reverse crossover. Use SMA 20 and 50.” This sample brief contains no further rules.
Method & assumptions
We review the specification for ambiguity. A small synthetic logic check also examines closed bars, warm-up, signal timing and independence from future data. It uses artificial values and shortened SMA windows of 2/3 only. It produces no market, return or robustness claims.
Findings & priorities
“Crossover” without evaluation timing
The deliberately incomplete sample brief does not specify whether the current or a closed bar counts. This can change signals retrospectively.
Next step: Evaluate closed bars only. Signal after close of t; earliest execution at the next tradable tick.
Missing risk model
No sizing, loss limit or restart definition. Automated orders cannot be approved on this basis.
Next step: Specify position limits, sizing, protective orders and failure state before execution testing.
Undefined costs and data
Without a source, spreads, fees and slippage, a test result is not reproducible. No such test has been supplied.
Next step: Create a versioned test manifest; record cost assumptions and data model before running it.
No independent evaluation period
Changing parameters after each evaluation reuses the same data for development.
Next step: Log attempted variants and reserve a predefined out-of-sample period.
Undefined operational behavior
Duplicate signals, disconnections, gaps and rejected orders are absent from the brief.
Next step: Specify idempotent signal handling, position reconciliation and error logging.
Proposed specification
Long-only. SMA 20/50 from fully closed bars; at least 51 historical closes before the first crossover comparison. At most one open position. Signal t: fast SMA previously at or below slow, now above. Exit on the reverse crossover. Earliest execution after the signal close. Market, timeframe, sizing, protective stop and operating rules must be completed before any backtest.
Backtest design & metrics
First freeze rules and parameters, document data and costs, and define development and untouched evaluation periods. Only then evaluate profit factor, win rate, average trade, equity drawdown, trade count and time distribution together. These values are explicitly unavailable here.
Bias and robustness
Synthetic signal checks cover selected logic cases; they do not establish market robustness. Parameter stability, independent periods, cost changes and forward behavior have not been tested. These missing checks require separate testing before deployment decisions.
Automation readiness
Not approved. The signal core can be described; a complete execution and risk model is missing. Next: resolve P0 gaps, then specify a reproducible baseline test.
Download sample report ↓Request a review of your strategy →
Demonstration, not investment advice. Historical tests and technically correct software do not guarantee profits.