Automated system for predicting cryptocurrency market trends

The challenge

High market volatility demanded adaptive models, and raw market data required complex feature engineering. The client also needed efficient parallelized data pipelines, a centralized system for testing multiple time series models, guaranteed data quality, and support for different forecasting schemas.

Our solution

• Time series forecasting solution using adaptive models tailored for volatile financial markets.
• Rigorous feature engineering to enhance prediction accuracy and capture market trends.
• Centralized system for testing and developing diverse forecasting models.
• Microservices architecture with admin UI, user UI and a customer API.
• Integrated Polygon API for historical crypto data, Google Cloud Vertex AI for model deployment and Azure Functions for post-processing.

Results

• The platform enables tracking of market trends and analysis of historical data.
• Users can create custom forecasting scenarios for portfolio diversification and return optimization.
• A suite of tools helps investors make informed investment decisions.