QM014 Hands-on Lab
Information Timing, Data Vintages, and Real-Time Macro Data
Objective
Classify macro observations at a fixed forecast cutoff as ADMISSIBLE, NOT_ADMISSIBLE, or UNRESOLVED using actual availability rather than the reference period or vintage label alone.
Run
From the QM014 folder:
python labs/python/qm014_information_timing.pyWhat to inspect
The three synthetic observations deliberately separate reference period, release time, vintage label, and actual availability.
- The May observation with a proven June 15 availability is admissible at the June 30 cutoff.
- The second May observation remains unresolved because actual availability is unknown.
- The June observation is not admissible merely because its vintage label says June; its release occurs after the cutoff.
Experiment
Move the cutoff earlier or later, or supply an actual availability time for the unresolved observation. Explain why changing a vintage label alone should not change the exact admissibility result when the availability evidence is unchanged.
Interpretation boundary
The lab illustrates a fail-closed information clock. Real datasets may require source-specific release calendars, time zones, revision histories, and conservative buffers when exact availability cannot be established.