About DIVA Quantizer
DIVA Quantizer
DIVA Quantizer is an educational portfolio analysis tool that packages academically validated quantitative models — Markowitz mean-variance optimization, Fama–French 3-/5-factor regressions, and López de Prado's Hierarchical Risk Parity (HRP) — into an interface that non-specialists can use.
Enter your holdings and we compute risk, expected return and factor exposures from historical price data — all in one screen. Every analysis is free and no signup is required.
Who builds it
DIVA Quantizer is a project run by a solo developer. It translates open academic research into code so that students of financial engineering and people thinking about their own asset allocation can use it for free. That said, the creator is not a finance professional or researcher, so please do not take the results at face value or apply them directly to your investments.
For questions, feedback or bug reports, please reach out at s1836373@gmail.com.
Models we offer
- Markowitz mean-variance optimization — efficient frontier, min-variance / max-Sharpe portfolios
- Fama–French 3-Factor — market (MKT), size (SMB) and value (HML) exposures with alpha
- Fama–French 5-Factor — extends the 3-factor model with profitability (RMW) and investment (CMA)
- HRP (Hierarchical Risk Parity) — hierarchical risk allocation that avoids inverting the covariance matrix
Data and analysis
Analyses use public market data and factors that are standard in the academic literature. US-stock factors come from a public library; Korean-stock factors we compute ourselves from public Korean market data. Prices and factor values are applied at analysis time, so all you have to provide is the tickers and amounts.
We do not guarantee the availability, accuracy or timeliness of external data sources, and every result is a statistical estimate based on historical data.
Tech stack
Backend: Python (Flask · pandas · statsmodels · numpy). Frontend: React · TypeScript · Vite. Result charts render entirely on the client. Holdings you enter are sent to the server only at analysis time and are not stored persistently.