Concepts
Concepts
Assumptions and Limitations
Canonical assumptions, constraints, and non-goals for model interpretation and backtest usage.
This page is the canonical reference for interpretation boundaries in Stacking Sats docs.
Core Assumptions
- Historical features and signal relationships remain informative enough for decision support.
- Data inputs are available, timely, and correctly transformed.
- Dynamic weighting preserves total budget constraints over evaluation windows.
Known Limitations
- Backtests are historical and do not guarantee future outperformance.
- Model outputs depend on feature quality and preprocessing decisions.
- Regime changes can reduce the relevance of historical calibration.
Non-Goals
- Predicting exact local tops or bottoms.
- Replacing user risk management or fiduciary advice.
- Guaranteeing better outcomes than uniform DCA in every period.
Operational Risk Checklist
- Confirm data and feature pipelines before relying on daily outputs.
- Review backtest provenance markers (
As of,Source artifact,Generation command) in Backtest Overview. - Use definitions from Glossary to avoid interpretation drift.