Documentation
Democratizing DeFi with Shwuan
Shwuan actively fetches blockchain data from our nodes,
accepts natural-language queries, and returns personalized, structured datasets for analysis and reports.
Data Agent · fetches, organizes, and computes blockchain data
Privacy · models are not trained on user queries
Sandbox · quantitative financial models
Typical query latency
< 1.5s
Time-windowed, single or pair assets
Supported metrics
4 core families
Spreads, z-scores, correlations, ratios
Shwuan is a DeFi agent layer between blockchain data and your analysis workflow.
You describe the dataset and analysis you want in layman's terms; Shwuan:
- Identifies which assets / CSVs you are referring to (e.g. btc, sol).
- Respects the time window in your prompt (e.g. “last month”, “2022 to 2024”, “past 2 years”).
- Organizes export files: statistics, formatting, strategies.
- Executes those operations and returns a downloadable table and chart.
Your queries are never used for model training. All user data is maintained in private infrastructure. We will never sell your data.
Time windows are resolved by the engine before any modeling. You can use natural language ranges:
- Relative windows: “last 30 days”, “past month”, “past 3 months”, “past 2 years”.
- Calendar ranges: “from 2022 to 2024”, “2023 only”.
- Recent shorthand: “recent”, “latest” (defaults to the last ~30 days).
Sampling:
- Daily – default, one row per trading day in your CSV.
- Weekly – say “weekly close prices” or “per week”; the engine resamples to end-of-week closes.
1. Describe the dataset. Examples:
- “BTC and SOL daily close prices over the past year, include spread and z-score columns.”
- “BTC and BNB monthly volume from 2022 to 2024 with a rolling 30-day correlation column.”
- “MetaX and Meta daily prices with the z-score of returns and differences (mean reversion stat arb.).”
2. Choose output. Select CSV, JSON, markdown, or text. The agent returns:
- A formatted table with dates, raw columns, and derived metrics.
- A single-series chart configuration (used by the UI to render the visualization).
3. Download & model. Export the CSV and plug it into Python, R, Excel, or the strategy lab.
