Quantitative Finance Research Papers

Volume-synchronized probability of informed trading as a practical signal for detecting adverse selection in fragmented equity markets. Includes Python implementation, volume bucketing, and trade classification methods.

Why GPU-accelerated financial modeling still requires CPU-class hardware. A split-path architecture analysis for real trading systems with latency decomposition and PyTorch inference examples.

Cluster-based portfolio allocation that avoids covariance inversion and delivers more stable out-of-sample diversification. Includes correlation distance, recursive bisection, and full Python implementation.

A statistically rigorous alternative to raw Sharpe that adjusts for non-normality, sample length, skewness, and kurtosis. Detects backtest overfitting with inferential statistics.

Quoted, effective, and realized spreads as microstructure state variables. Decomposing the cost of immediacy for execution models with adverse selection analysis.

Why self-hosted language models are structurally superior for investment research. The Bastion philosophy — data minimization, auditability, latency predictability, and customization.

Evolutionary search as an alpha hypothesis generator. Fitness design, selection pressure, and rigorous out-of-sample validation for non-convex trading strategy spaces.

Why PnL arises from signal after implementation, not signal alone. Fill models, latency decomposition, square-root market impact, and Python fill simulators.

Satellite imagery, transaction panels, and e-commerce data as nowcasting signals. Data governance, timestamp integrity, and ML pipeline design for alternative data sources.

Rolling-window mathematics, GPU-accelerated analytics, and feature engineering for quantitative trading systems. From raw market data to predictive feature matrices.

Building a time-aware, Turkish-native NLP pipeline using Qwen and Llama for financial news signal extraction on Borsa İstanbul. Local inference, label design, and critical pitfalls.

From a $14,000 Polymarket return to Goldman Sachs automation and a trillion-dollar tech selloff — quantitative analysis of how Claude AI has become the most disruptive force in financial markets since algorithmic trading.