This repo stores the project accomplished following the university course of Financial Risk Management held by Prof. Cosimo Munari & Prof. Andrea Mazzon, at the University of Verona, within the Master's Degree in Quantitative Finance.

The original instructions can be found in Lecture 11 Java Project.pdf.

SUMMARY: Java implementation of an n-asset portfolio risk management system computing Value at Risk (VaR, α=1%) and Expected Shortfall (ES, β=2.5%) via three estimation methods — Historical (non-parametric empirical quantile), Normal parametric (Gaussian fit), and Monte Carlo with s simulations per rolling window (user-configurable in Tests.java), in both independent and correlated (Cholesky decomposition) variants. The framework is generalized to any number of assets: portfolio weights are derived from an arbitrary budget vector (BUDGETS), and the full Historical / Normal / Independent MC pipeline scales to n assets without modification. The correlated MC variant is the only component restricted to two assets, relying on a closed-form 2×2 Cholesky factorization. The demo portfolio uses SPY (S&P 500 ETF) and GLD (Gold ETF), observed over 2019–2025 with a 250-day rolling window, with results expressed both as percentage of portfolio value and in USD. The project also includes Basel III Traffic Light backtesting evaluated annually (2020–2025) and on a rolling 250-day basis. Key findings: the Historical method is the most conservative and never enters the RED zone; the Normal and Correlated MC methods underestimate fat-tail risk, producing RED violations in 2022 and 2025; the Independent MC model — which ignores the negative SPY–GLD correlation — yields the worst backtesting performance, confirming the diversification benefit of including gold as a hedge. The codebase follows the Strategy, Facade, and Template Method design patterns.

*Place the historical price files in src/main/resources/ as Asset1.xlsx, Asset2.xlsx, etc. Files must be in the standard Yahoo Finance download format (.xlsx): the Date column in the first position and an Adj Close (or Close) column containing the adjusted closing prices. One file per asset, one row per trading day.