Crypto Basket Predictor

Web platform developed during a hackathon to predict and visualize cryptocurrency basket price trends using time-series machine learning models, Django, Vue.js, and MySQL.

Vue.jsDjangoMySQLMachine LearningTime-SeriesHackathon
View on GitHub

Overview & Hackathon Challenge

The Crypto Basket Predictor was built during the fast-paced Bugsbyte Hackathon by team GourmetRace. The goal of the project was to create a full-stack platform capable of tracking, weighting, and forecasting the combined price movement of customized cryptocurrency "baskets" (similar to digital asset index funds or ETFs) rather than evaluating volatile coins in isolation.

At the time of the hackathon, I was still building my foundation across many core concepts in full-stack architecture, API design, and machine learning workflows. Stepping into a competitive, time-crunched hackathon environment forced me to step far outside my comfort zone. I had to rapidly read documentation, grasp unfamiliar paradigms in hours, adapt on the fly, and ship working features under strict deadlines alongside my teammates.

Crypto Basket Predictor

The Problem: Navigating Crypto Volatility

Individual cryptocurrencies are notoriously volatile, with extreme price swings driven by market sentiment and liquidity shifts. For retail investors and analysts, examining single assets in isolation often gives a distorted view of broader sector trends (such as DeFi, Layer 1 blockchains, or payment tokens).

Our solution introduced thematic baskets:

  • Users can compose weighted portfolios of multiple digital assets (e.g., Bitcoin, Ethereum, and selected altcoins).
  • The platform calculates composite performance metrics, smoothing out idiosyncratic noise.
  • Automated Time-Series Machine Learning models analyze historical price dynamics and project future basket trajectory trends to support informed decision-making.

Architecture & Technical Stack

Despite the rapid development timeline, the team established a clean separation of concerns across the stack:

1. Frontend Client (Vue.js)

  • Interactive UI: An intuitive interface where users can select individual cryptocurrencies, adjust percentage allocations within their custom basket, and inspect historical performance.
  • Dynamic Data Visualization: Integrated client-side charting to display synchronized price curves, comparing actual historical prices against model-predicted trend lines.
  • State Management: Handled responsive UI updates as users rebalanced basket weights in real time.

2. Backend & ML Engine (Django & Python)

  • RESTful API: Built with Django to expose endpoints for basket configuration, cryptocurrency market data queries, and prediction requests.
  • Data Ingestion Pipeline: Ingested historical Open-High-Low-Close-Volume (OHLCV) market data, transforming raw payloads into structured time-series datasets.
  • Forecasting Pipeline: Applied preprocessing steps (normalization, rolling windows, and feature scaling) to feed open-source time-series models, computing composite projections based on the user's asset weights.

3. Database Layer (MySQL)

  • Relational Modeling: Stored cryptocurrency metadata, historical pricing snapshots, user basket definitions, and cached prediction outputs to minimize redundant external API requests.