Quantitative backtests and portfolio optimization in Python and R, and two finance web apps used to analyse markets and learn finance. I care about results that hold up out of sample.
Two web apps I designed and launched, and two quantitative research projects.
Market analysis for non-specialists: every figure comes with its definition and a one-line interpretation.
Stocks, ETFs, portfolios (efficient frontier, risk contributions), a 2,600-asset screener, bonds, options and an economic calendar. French and English.
Built with AI-assisted coding. First load can take a few seconds.
Learn finance a few minutes a day, in the spirit of Mimo and Codecademy.
186 lessons in French and English across 12 topics (markets, accounting, derivatives, quant…), 5 levels each, with exercises, streaks and XP.
Built with AI-assisted coding.
A gross Sharpe ratio above 4 that disappears out of sample once a 1-hour execution lag and trading costs are included.
Cointegration (ADF), Ornstein-Uhlenbeck modelling, EWMA signal bands, return/CVaR allocation, and a backtest with AMM fees, slippage and gas, on a strict train/test split.
On 99 Nasdaq-100 stocks, minimum-variance cut volatility by a quarter out of sample, but no portfolio beat the index.
Ledoit-Wolf covariance, minimum-variance, index tracking and efficient frontier, cross-checked with two solvers (quadprog, nloptr) and evaluated over 25 unseen months.
Final-year student in financial markets, looking for an end-of-studies internship in 2027.
Internship opportunities, questions about a project, or feedback on the apps.