Purpose and protocol
The dossier is an educational test of reproducibility. Its fictional universe and constructed returns support checks of ranking, timing and transaction-cost accounting. No security is recommended and no actual instrument model is trained. Dates in a synthetic fixture describe the example clock, not a historical observation record.
Information boundaries
A score must have an availability timestamp strictly before the stated formation cut-off. Selection uses only scores admitted before that cut-off. A current vendor download is not automatically point-in-time data: historical revisions require their own vintages and timestamps. Corporate-action conventions and universe inclusion decisions belong in the input record.
Sample discipline
The fixture identifies design, validation and holdout periods. These labels demonstrate a protocol; such a short synthetic sequence cannot estimate investment skill. Changing settings after viewing results is sensitivity analysis. It does not create a fresh independent test.
Reproduction and limits
Download the fixture, run the published engine and compare the period ledger with the independent reference calculations. The engine states its rebalance and stock-cost conventions. It does not model order-book fills, taxes, financing, borrow, market impact or production execution. Unit tests establish arithmetic behavior, not out-of-sample profitability.