Python Forex Pattern Analysis
Budget: $30 – $250 USD
I have access to 10–20 years of historical tick and/or OHLC data and I want to put it to work. The goal is clear: build a Python-based workflow that scans that long-term price history, flags statistically significant reversal patterns, and produces a concise report I can rely on for future strategy design.
Scope
• Import and clean multi-decade forex data (I can provide raw CSVs or you can script an automatic pull from common sources such as Dukascopy, OANDA or HistData).
• Run exploratory analysis to normalise sessions, handle weekend gaps and adjust for rollovers.
• Implement robust pattern-detection logic focussed on reversals—think double tops/bottoms, head-and-shoulders, evening and morning stars, hammer / shooting-star clusters, plus any quantitative approach such as mean-reversion signals based on z-scores or Bollinger touches.
• Back-test each detected pattern across all major pairs I supply, then output win rate, R-multiple distribution and maximum drawdown so I can quickly judge viability.
• Package everything in readable Jupyter notebooks and a lightweight CLI script so I can rerun it on fresh data.
Key tech
Python 3.x, pandas, NumPy, SciPy, TA-Lib (or a custom implementation if you prefer), scikit-learn for any statistical tests, and Matplotlib/Plotly for visual summaries.
Timeline
Deliver a first working draft inside two weeks so there is room for refinement before the one-month window closes.
Deliverables
1. Clean, well-commented source code (Git-ready).
2. Jupyter notebook walkthrough with charts, tables and interpretation notes.
3. A brief PDF or Markdown report summarising findings and suggested next steps.
Acceptance
Code must run end-to-end on my machine with a single command, reproduce the sample figures you show in the notebook, and correctly identify at least the classic reversal setups on a validation slice of data I withhold until handover.
If you have previous experience mining long-horizon forex datasets or have proprietary pattern libraries you’re willing to adapt, I’d love to see your approach.
Scope
• Import and clean multi-decade forex data (I can provide raw CSVs or you can script an automatic pull from common sources such as Dukascopy, OANDA or HistData).
• Run exploratory analysis to normalise sessions, handle weekend gaps and adjust for rollovers.
• Implement robust pattern-detection logic focussed on reversals—think double tops/bottoms, head-and-shoulders, evening and morning stars, hammer / shooting-star clusters, plus any quantitative approach such as mean-reversion signals based on z-scores or Bollinger touches.
• Back-test each detected pattern across all major pairs I supply, then output win rate, R-multiple distribution and maximum drawdown so I can quickly judge viability.
• Package everything in readable Jupyter notebooks and a lightweight CLI script so I can rerun it on fresh data.
Key tech
Python 3.x, pandas, NumPy, SciPy, TA-Lib (or a custom implementation if you prefer), scikit-learn for any statistical tests, and Matplotlib/Plotly for visual summaries.
Timeline
Deliver a first working draft inside two weeks so there is room for refinement before the one-month window closes.
Deliverables
1. Clean, well-commented source code (Git-ready).
2. Jupyter notebook walkthrough with charts, tables and interpretation notes.
3. A brief PDF or Markdown report summarising findings and suggested next steps.
Acceptance
Code must run end-to-end on my machine with a single command, reproduce the sample figures you show in the notebook, and correctly identify at least the classic reversal setups on a validation slice of data I withhold until handover.
If you have previous experience mining long-horizon forex datasets or have proprietary pattern libraries you’re willing to adapt, I’d love to see your approach.
Related categories:
Python
Software Architecture
Statistics
Statistical Analysis
NumPy
SciPy
Data Visualization
Data Analysis