Skip to system research
Build A Trading SystemRules / Evidence / Replay

Synthetic fixtures and point-in-time market history

Use invented data for software tests without laundering it into a historical trading source.

The source category is a field, not an impression

The reference CSV is deliberately constructed data with arbitrary UTC timestamps. No exchange, licensed feed, real ticker, corporate action or historical universe supplied it. Its completed shape makes it useful for deterministic software tests, not for measuring a market hypothesis. A familiar date format or professional chart does not change the classification. Keep invented status in the CSV description, replay receipt, visible workbook and machine records. Never substitute plausible synthetic rows for missing historical data and report them as observed evidence because the downstream simulator accepts the file.

Historical identity includes availability and revisions

A real data contract needs instrument identity, producing source, observation and availability times, timezone, relevant revisions, normalization and point-in-time membership. Current survivors cannot automatically describe an older universe, and a revised release may not be what a decision could originally know. Individual futures contracts and separately labelled continuous research references also have different scope. A toy file containing generic prices does not resolve those routes. The authorised instrument process must select and verify the actual source before a stronger financial conclusion, without erasing identity guards or treating unavailable individual contracts as a universal blocker to reference research.

Synthetic defects can still teach useful behaviour

The small loader rejects duplicate IDs, duplicate times, invalid OHLC ranges, nonfinite values and missing rows. Its input bars are strictly ordered one-minute observations under an invented clock. These checks demonstrate selected parsing and identity behaviour. They do not establish that a real feed has no gaps, correct corporate actions or permissible redistribution. The prefix mutations change later invented observations to check causal processing; they are not historical revisions or independent validation periods. Retain the exact source bytes and contract so a reader can distinguish a completed software case from an absent market source.

A completed negative research boundary

The frozen rule produces six simulated positions and negative 1.20 net cash under the declared fee policy. That result is complete as a software reference but cannot accept or reject a real market family. A verified no-model master requires the separate complete authorised research and independent checks, not a synthetic example with a negative total. Likewise, a positive variation would not be validated profit. The publication does not promote this fixture, replace canonical fill conventions or invent raw-data rights. The useful teaching conclusion states exactly which software questions are answered and which economic questions still need actual producing evidence.

Filled invented system case

The fictional builder receives a CSV with 64 invented OHLC bars and a working replay launcher. The loader and prefix tests pass, but the source category stays synthetic. No point-in-time real universe, exchange-session clock or independent market sample exists. The recorded negative 1.20 belongs to the library fixture only. Renaming its arbitrary instrument field to a real ticker would not create historical data, suitable calibration, independent strategy evidence or a certified research result.

ObjectRecordBounded finding
SourceDeliberately invented barsSoftware fixture
ClockArbitrary one-minute UTC labelsNo exchange session
Input checksIDs / times / OHLC / finite / countParsing and contract tests
Economic validationActual market source absentNot performed
Invented closes and moving averages above simulated equity
Invented library replay, not historical data or economic validation.

Primary references and scope

Expanded source review: 13 October 2026. Documentation supports the stated mechanism, not a tested account, hypothetical fill or provider outcome.

Related research questions