# Wansiri Research Platform ## Overview The Wansiri Research Platform is organised as a series of independent research pipelines. Each pipeline has a defined responsibility and produces its own database. These databases are reused by subsequent pipelines, backtesting, reporting and, ultimately, live trading agents. The architecture is designed so that research outputs are reproducible, auditable and based only on information available at the time being evaluated. ## Research Pipeline Architecture 1. Raw Market Data 2. Market Structure Pipeline 3. Feature Pipeline 4. Backtesting Pipeline Each component operates independently and produces data that can be consumed by subsequent research processes. ## 1. Raw Market Database **Purpose:** Store the historical market data used by the platform. It is the source of truth for research and historical analysis. **Questions it answers:** - What happened in the market? - What information was available at a given point in time? - Can a research result be reproduced from the underlying data? ## 2. Market Structure Database **Purpose:** Describe how market structure evolves over time, independent of any particular trading strategy. Metrics are calculated using only information available at the relevant point in time. **Example contents:** - Relationship metrics - Market condition metrics **Questions it answers:** - How is market structure evolving? - What information was available at a specific point in time? ## 3. Feature Database **Purpose:** Generate research features for market analysis, prediction or trading, developed independently of any specific strategy wherever possible. **Example contents:** - Statistical, price-based and volume-based features - Custom engineered features **Questions it answers:** - Which features are robust and stable through time? ## 4. Backtesting Database **Purpose:** Evaluate trading strategies using the research databases, and keep a structured record of the results. **Example contents:** - Trade and order records - Performance and robustness assessments ## Research Platform Consumers Research outputs are made available to downstream processes, including strategy research, portfolio construction, risk management, reporting and live trading agents. This allows outputs to be reused without every consumer repeating the underlying calculations. ## Design Principles - **Single responsibility:** Each pipeline has a defined purpose. - **Independent execution:** Pipelines can operate independently. - **Reproducibility:** Results can be reproduced from their underlying data. - **Auditability:** Research outputs are traceable. - **Temporal integrity:** Calculations use only information available at the relevant time. - **No look-ahead bias:** Future information never influences historical decisions. - **Reusability:** Outputs are available to multiple downstream processes. - **Strategy independence:** Market structure and feature research remain independent of individual strategies wherever possible. ## Relationship to Live Trading Historical research and backtesting help evaluate strategies before production consideration. Live trading agents consume relevant research outputs and operate against current market conditions. The research platform and the live trading environment are connected but distinct parts of the Wansiri system. ## Summary The platform separates market data, market structure analysis, feature generation and strategy evaluation into independent pipelines. This supports reproducibility, auditability, temporal integrity and reuse of research outputs across the wider platform.