Volatility Contraction Pattern Algo Development
Budget: ₹1,500 – ₹12,500 INR
I want a clean, well-documented Python routine that reliably spots the Volatility Contraction Pattern (VCP) in equities. The code must be able to ingest both historical price series (for back-testing) and a live market feed (for real-time screening), then return a clear signal whenever a qualifying pattern finishes forming.
Key points you should factor in as you build:
• Data scope – daily and intraday OHLCV from yahoo finance or a zerodha broker/API stream; your code should let me swap data sources with minimal edits.
• Pattern logic – I follow Mark Minervini’s classic VCP definition (series of narrowing price/volume contractions that resolve near a pivot). Please encode that logic with tunable hyper-parameters so I can tighten or loosen the rules later.
• Tech stack – pure Python, leveraging pandas, NumPy, TA-Lib (or your own study functions), and websockets/REST wrappers for the real-time side.
• Output – a simple, structured signal object (ticker, time stamp, stop level, entry level) that I can pick up in another script. Optional helper functions for quick Matplotlib plots are welcome for visual validation.
Deliverables
1. A modular .py package (back-test module, real-time module, shared utilities).
2. Example Jupyter notebook showing back-test results on at least one major index constituent.
3. README with setup instructions and parameter explanations.
4. Short demo clip or GIF proving the real-time alert fires as expected.
Acceptance Criteria
• Reproduces historical VCPs on test tickers with <2 % false-positive rate in sample period I supply.
• Able to run continuously on a live feed for at least one trading session without crashing or memory leaks.
• Code passes pylint 8.0+ score of 8 or higher and includes docstrings/type hints.
If you have previous work detecting chart patterns or streaming market-data projects, great—send a one-liner summary with your bid. Let’s keep the focus tight: functional, transparent, and ready for me to extend once it’s in place.
Key points you should factor in as you build:
• Data scope – daily and intraday OHLCV from yahoo finance or a zerodha broker/API stream; your code should let me swap data sources with minimal edits.
• Pattern logic – I follow Mark Minervini’s classic VCP definition (series of narrowing price/volume contractions that resolve near a pivot). Please encode that logic with tunable hyper-parameters so I can tighten or loosen the rules later.
• Tech stack – pure Python, leveraging pandas, NumPy, TA-Lib (or your own study functions), and websockets/REST wrappers for the real-time side.
• Output – a simple, structured signal object (ticker, time stamp, stop level, entry level) that I can pick up in another script. Optional helper functions for quick Matplotlib plots are welcome for visual validation.
Deliverables
1. A modular .py package (back-test module, real-time module, shared utilities).
2. Example Jupyter notebook showing back-test results on at least one major index constituent.
3. README with setup instructions and parameter explanations.
4. Short demo clip or GIF proving the real-time alert fires as expected.
Acceptance Criteria
• Reproduces historical VCPs on test tickers with <2 % false-positive rate in sample period I supply.
• Able to run continuously on a live feed for at least one trading session without crashing or memory leaks.
• Code passes pylint 8.0+ score of 8 or higher and includes docstrings/type hints.
If you have previous work detecting chart patterns or streaming market-data projects, great—send a one-liner summary with your bid. Let’s keep the focus tight: functional, transparent, and ready for me to extend once it’s in place.