- Home
- Skills
- Trading & Investing
- stratproof lab
stratproof lab
Deterministic crypto strategy backtesting and audit engine for verifiable trading hypothesis research.
Free
Works with the AI Agents you already use
stratproof lab
Example session with this skill installed
Audit a LONG strategy for SOLUSDT on 5m candles. Rules: RSI < 30 and Price > EMA 50 on BTCUSDT 15m. Use 0.5% TP and 0.5% SL over the last 30 days of Bybit data.
- Read your context and instructions
- Compiled the stratproof lab
I have completed the deterministic audit for the requested SOLUSDT LONG strategy. The engine replayed 30 days of Bybit historical data using a 12-candle horizon for outcome evaluation.
Strategy Audit Report: RSI_BTC_EMA_LONG
| Metric | Value |
|---|---|
| Status | IDEA_BACKTEST_COMPLETE_AUDIT_ONLY |
| Idea Hash | f7a2b9e1c8d3... |
| Total Signals | 42 |
| Closed Outcomes | 38 |
| Win Rate | 57.89% |
| Net R (Simple) | +6.0 |
| Verdict | POSITIVE_EXPECTANCY |
Indicator Performance Breakdown
The strategy required an RSI < 30 on the 5m timeframe synchronized with BTCUSDT trading above its 15m EMA 50. Signals were most frequent during the London-New York overlap. 4 signals remain "Open" as they did not hit TP/SL within the 12-candle horizon.
Evidence Artifacts
- JSON:
reports/idea_lab/f7a2b9e1c8d3.json - Markdown:
reports/idea_lab/f7a2b9e1c8d3.md
Next steps
- Review the
reports/idea_lab/f7a2b9e1c8d3.mdfile for a candle-by-candle log of every signal. - Run
stage13_run_multitimeframe_audit.pyto see how varying the RSI threshold (e.g., 25 vs 35) affects the Net R. - Note: This audit does not account for 0.06% taker fees, which would reduce the Net R significantly.
Connects securely to your tools. The creator never sees your data.
About this skill
Manual backtesting is slow, error-prone, and lacks verifiable evidence. StratProof Lab provides a deterministic audit engine to replay rule-based cryptocurrency trading hypotheses against historical exchange data.
What it does
- Multi-exchange integration fetches public OHLCV data from Bybit, Binance, OKX, Coinbase, and Kraken.
- Rule-based evaluation tests strategies using RSI, EMA, BTC market regime context, and relative volume filters.
- Session filtering isolates performance during specific windows like the New York Open.
- Audit trail generation produces verifiable JSON, Markdown, and CSV evidence artifacts for every run.
- Performance reporting calculates win rates and simple Net R while handling TP/SL ambiguity conservatively.
How it works
- Stage data by downloading historical candles via
stage11_download_history.pyor generating synthetic sets for testing. - Define the recipe using a JSON schema to specify indicators, operators, thresholds, and symbols.
- Execute the audit using
stage12_run_idea_backtest.pywith custom horizons and risk parameters. - Analyze findings via the generated Markdown reports and structured CLI output.
Frameworks & tools
Built with Python 3 using only the standard library. Supports Bybit, Binance, OKX, Coinbase, and Kraken.
Why this beats prompting it yourself
Prompting an LLM for backtests results in "hallucinated" performance data and math errors. This skill runs deterministic code against actual historical candles, ensuring your strategy validation is based on math rather than vibes.
Use cases
- Validate RSI oversold strategies paired with high-timeframe BTC trend filters.
- Audit the impact of relative volume spikes on breakout success rates.
- Compare strategy performance across multiple symbols like SOL, ETH, and BTC.
- Generate evidence reports for trading hypotheses before moving to production.
Known limitations
Does not model exchange commissions, slippage, or funding fees. Requires local or cached OHLCV data to function.
No modeling of commissions, slippage, or funding fees. Research-only; no live execution or API key support. Requires manual download of OHLCV data before auditing.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed today
- Free with an Agensi account
Requires
Permissions
Allowed Hosts