SYNTHETIC FIXTURES / OFFLINE REPLAY
Evidence, before
confidence.
A good system can explain why it trusts a result.
Run three small checks that make the reasoning visible.
03 / CHECKS
No market connection.
No account data.
Every result is reproducible.
THE TEST BENCH01 / 03
01 / TEMPORAL INTEGRITY
A decision at 2 seconds
cannot see 4 seconds.
An event arriving later can change the full recording. It must never rewrite what the system knew at decision time.
THE INVARIANT
Use only events with a timestamp at or before the decision.
ASSERTION LOGREADY TO RUN
| CHECK | EXPECTED | OBSERVED | Status |
|---|
Run the fixture to inspect the expected and observed values.
Small fixtures. Explicit limits. These checks demonstrate engineering behavior with invented inputs. They are not market returns, a strategy backtest, or evidence of future profitability. The bar example uses a conservative assumption because OHLC data cannot reveal the order of events inside a candle.
READ THE FIXTURE NOTES