Selected work
04 · Probabilistic portfolio research
AI Tech Market Simulator
A six-month scenario engine for reviewing a fixed AI-technology portfolio through an institutional dashboard lens.
six-month scenario enveloperesearch simulation
BaseBullBear
Executive summary
A Streamlit MVP that assembles market history, transparent event rules, technical indicators, Monte Carlo paths, and portfolio-level risk views without claiming exact price prediction.
Problem
A market research interface needs to make uncertainty legible instead of presenting a single deterministic forecast as fact.
Approach
- 01Load six months of OHLCV history with deterministic fallback data when an external market-data request fails.
- 02Calculate returns, rolling volatility, moving averages, RSI, MACD, Bollinger Bands, and volume trends.
- 03Classify typed event text with transparent positive, negative, mixed, or neutral keyword rules.
- 04Generate base, bull, and bear Monte Carlo paths, then summarize value ranges and risk metrics.
02
Data & methodology
Data sources
- yfinance market data
- Fixed project portfolio weights
- Deterministic fallback data for offline resilience
Methods
- Technical indicators
- Rule-based event classification
- Monte Carlo scenario paths
- Volatility, drawdown, and Sharpe-ratio summaries
Workflow
Input→Collection→Analysis→Output
The project README explicitly frames forecasts as scenarios and not guaranteed outcomes.
Outputs & findings
Key outputs
- Portfolio overview
- Candlestick and indicator views
- Scenario envelope
- Risk and benchmark charts
What the work demonstrates
- Delivered an end-to-end research dashboard that makes scenario assumptions visible.
- Keeps live-data failure from breaking the interface by clearly using deterministic fallback data.
Limitations
- A probabilistic scenario tool, not a price-prediction engine or investment advice.
- Version 1 uses transparent keyword rules rather than an LLM or a trained forecasting model.
- The portfolio is fixed and not optimized for a specific investor.