Signal Design Principles
Snowtrail Signals are built following consistent design principles to ensure reliability and usability.
Core Principles
1. Fundamental Grounding
Signals are derived from observable fundamental data, not from price patterns or technical indicators. This ensures signals capture real-world supply and demand dynamics.
2. Point-in-Time Construction
Signals are computed using only information that was available at the time. This prevents look-ahead bias and ensures signals can be reliably backtested.
3. Consistent Methodology
Signal definitions are stable over time. When methodology changes are necessary, the model_version column increments and the change is recorded in the changelog. See Versioning.
4. Interpretability
Each signal has a clear economic interpretation. Users should understand what the signal measures and why it might be predictive.
Normalisation
Signals are typically normalised to enable:
- Comparison across different time periods
- Comparison across related markets
- Combination with other signals
Update Frequency
Signals are updated when new underlying data becomes available, with clear documentation of the update schedule for each signal.