MarketWiki BD
Retail investors in Bangladesh's DSE/CSE market have no easy way to spot pump-and-dump schemes or gauge how trustworthy an issuer is before investing. MarketWiki BD is a market-intelligence platform I'm building to score issuer credibility and flag manipulation patterns automatically.
Bangladesh's stock market (DSE/CSE) has a long history of syndicate-driven pump-and-dump schemes and opaque issuer practices, and retail investors have no accessible tool to see the warning signs before they're already holding the bag. MarketWiki BD is my answer to that gap — a surveillance and scoring platform that turns public market data and filings into an early-warning signal.
What it does
- Computes an Issuer Trust/Authenticity Score from market and disclosure data.
- Classifies manipulation phases and fingerprints syndicate trading patterns.
- Flags influencer-driven pump activity and tracks political/regulatory risk signals.
Technology
Django REST Framework backend with JWT/API-key auth and drf-spectacular for the API layer; Channels/Daphne and Celery/RabbitMQ for real-time and background processing; PostgreSQL today, with TimescaleDB, Neo4j and a vector store planned for time-series and relationship analysis. Market data comes in through my own stocksurferbd library, with pandas for analysis and LangChain (Anthropic/OpenAI/Gemini) for document parsing and pattern reasoning. Stripe handles billing; the whole stack is Dockerized and deployed on Railway.
Status: pre-MVP — the product design and core infrastructure (auth, billing, data ingestion scaffolding) are in place; the manipulation-detection models are what I'm building next.
