Daris Fadhilah

An evidence-graded signal platform for retail investors. It compiles an investment philosophy into versioned config, ranks systematic research against it, and grades its own signals against realized forward returns.

• Options-wheel screener ranking cash-secured puts and covered calls by annualised return on collateral
• Append-only signal spine so every recommendation can be scored later, including published null results
• Layered architecture enforced with import-linter and mypy --strict; 75+ architecture decision records
• LangGraph and Pydantic AI research agents that run inside the same contracts as the rest of the code

• Python 3.13
• FastAPI
• PostgreSQL
• Polars
• DuckDB
• React
• TypeScript
• Docker

Calibration and hedging performance of eight option pricing model families on 0DTE to 30DTE SPX and single-name options, plus a Bitcoin volatility surface fitted with SVI.

• Models benchmarked: ad-hoc Black-Scholes, SVI, Heston, Merton, Bates, Variance Gamma, NIG and Edgeworth expansions
• Delta and delta-vega hedging backtests across maturity buckets
• Ported a collapsed Kalman filter for characteristic-function estimation of stochastic volatility models from Julia to Python

• Python
• NumPy
• SciPy
• QuantLib
• Julia

Quantitative research project with APG Asset Management and Erasmus University Rotterdam on unsmoothing reported hedge fund and private-market returns.

• Proposed a three-step regime-switching methodology that lowered variance in prediction errors across AR- and MA-based alternatives
• Benchmarked Geltner and Getmansky-Lo-Makarov unsmoothing on WRDS and TASS data
• Team of four; I led the codebase and the empirical pipeline

• Python
• statsmodels
• R
• WRDS