Data Model Overview
Snowtrail organises data into four core types: Signals, Features, Events, and Backtests.
Core Data Types
Signals
Quantitative classifications of market conditions, derived from fundamental data. Signals describe the state the market is in. They are built for position sizing and risk conditioning, not for predicting price direction.
Features
Curated, transformed datasets created from raw fundamental inputs. Features are designed as reliable building blocks for quantitative research and analysis.
Events
Structured records of discrete market developments such as outages, policy changes, or supply disruptions that may impact prices or fundamentals.
Backtests
Pre-computed, point-in-time correct performance analysis for each signal. Backtests measure whether regime classifications correspond to meaningfully different market outcomes.
Common Principles
All Snowtrail data types share these characteristics:
- Point-in-time integrity - Values reflect what was known at the time
- Consistent definitions - Clear, documented methodology
- Research-ready format - Structured for immediate use in analysis