Oceanic Edge is a quantitative research platform that identifies market-moving signals by combining maritime AIS (Automated Identification System) data with Satellite Computer Vision. The framework tracks ship congestion and port throughput to predict revenue outcomes for global logistics and retail giants.
This project demonstrates expertise in Spatiotemporal Data Analysis, Computer Vision, and Quantitative Alpha Generation.
- Spatiotemporal Geofencing Engine: Automatic detection of ship arrivals/departures at port anchorages using
Shapelypolygons. - Port Vision Engine (YOLOv8): Wrappers for YOLOv8 models designed to detect container stacks and ships in port terminals via satellite imagery.
- Dwell Time Calculation: Aggregation of continuous waiting periods for ships within port geofences using AIS telemetry.
- Port Congestion Index (PCI): Z-scored, rolling window signal representing severe shipping bottlenecks.
- Logistics-Retail Hedge Engine: Cross-sectional alpha correlation that generates Long/Short target weights based on supply chain disruptions.
- Language: Python 3.12
- Data Engineering: Pandas, NumPy
- Spatial Analysis: Shapely, GeoJSON
- Computer Vision: Ultralytics (YOLOv8)
- Visualization: Matplotlib, Seaborn
oceanic_edge/
├── data/ # AIS parsers and Port Geofence Registry JSON
├── spatial/ # Shapely geofencer point-in-polygon logic
├── models/ # Dwell Time engine and YOLOv8 Vision wrappers
├── signals/ # Port Congestion Index and ticker correlation
├── backtest/ # Hedging strategy target weight generators
├── notebooks/ # Research, analysis, and visualization
└── requirements.txt # Project dependencies
- Clone the Repository
git clone https://github.com/itzsam-lol/Oceanic-Edge.git
cd Oceanic-Edge- Configure Virtual Environment
python -m venv venv
.\venv\Scripts\Activate.ps1- Install Dependencies
pip install -r requirements.txtTo run the automated research flow that processes mock AIS data, calculates dwell times, and generates a target portfolio:
python pipeline.pyExplore the analytical walkthrough and visualization in the Jupyter Notebook:
- Congestion Alpha Analysis: View the logic behind extracting Z-score congestion signals and mapping them into trading weights.
- Geospatial Engineering: Overcame challenges with handling spatiotemporal data and efficiently mapping raw GPS pings to geometric boundaries.
- Signal Normalization: Modeled the financial impact of physical supply chain constraints via standardized Z-scores.
- Future Improvements:
- Integration with real-time AIS web sockets or APIs (e.g., MarineTraffic).
- Expanding the Port Registry to cover the top 100 global ports and mapping them to a wider universe of equities via GICS codes.