FinLove
An intelligent portfolio construction and analysis platform combining quantitative finance, forecasting, and retrieval-augmented generation.
FinLove is a full-stack portfolio management platform for investment planning, risk analysis, backtesting, and forecasting.
The platform combines portfolio optimization methods—including Markowitz, Black–Litterman, and CVaR—with risk models such as Ledoit–Wolf, GLASSO, GARCH, and DCC. Its forecasting pipeline evaluates ARIMA, Prophet, LSTM, and statistical models, then selects the strongest models to estimate future portfolio returns with uncertainty.
I also integrated an LLM-powered retrieval-augmented generation system to provide context-aware explanations of portfolio performance and risk metrics.
Stack: FastAPI, Next.js, Python, CVXPY, TensorFlow, and LLM-based RAG.