📚 Research Blog

In-depth articles on quantitative finance, high-frequency algorithms, and cutting-edge technologies (Rust, PyO3)

⭐ Featured · NEW log E[(Δlogσ)²] vs log Δ Brownian (H=0.5) H ≈ 0.114 (VIX/SPY) Rough Volterra + McKean–Vlasov + CVaR CVaR₉₉: $62 (vs $41 Brownian)
Optimiz-rs Volterra McKean–Vlasov Microstructure May 14, 2026

Physics-inspired microstructure: Volterra, McKean–Vlasov & risk measures with Optimiz-rs

Four v2 primitives composed on real datasets: a Volterra fit on VIX/SPY recovers $H \approx 0.114$, a propagation-of-chaos rate scaling as $N^{-1/2}$ on 2 h of BTCUSDT ticks, and 99 % VaR / CVaR computed natively in Rust. The Brownian baseline under-estimates CVaR by 30 %; the physics-inspired pipeline matches the empirical tape to within 5 %.

⭐ Featured · NEW Superspace → minisuperspace dim ~ N×T BRST / Wheeler-DeWitt d_eff = 4 Dirac op. · η-invariant · H₁
Mathematical Physics McKean-Vlasov Topology Crypto · ETF May 24, 2026

Taming Chaotic Markets — Physics-Inspired Models from McKean-Vlasov to Topological Field Theory

Mean-field games, the Wheeler-DeWitt minisuperspace ansatz, BRST supersymmetry on ETF/crypto baskets and the Dirac market operator. On 7 real underlyings, the effective dimension drops to $d_{eff}=4$ and the η-invariant detects regime breaks 4-9 days ahead. A public preview of the building blocks behind the private WP-5 whitepaper.

⭐ Featured · NEW E[PnL] with / without option overlay 0 Without overlay -$8.69 + Gamma harvest +$3.42 Sharpe -1.29 → +0.74 (PCA basket)
Arbitrage Multi-asset OU Riccati Options 26 April 2026

Cross-asset optimal execution under Ornstein–Uhlenbeck dynamics

Bergault–Drissi–Guéant 2022 + Hawkes-adaptive impact + 0–7 DTE option overlay. Real Binance/Deribit KPIs: $E[\text{PnL}]$ flips from $-\$8.69$ to $+\$3.42$ with gamma harvest; PCA basket Sharpe $-1.29 \to +0.74$. Comes with the 17-page commercial course and a free PDF primer (Riccati, Hawkes, viscosity, convex analysis).

⭐ Featured Queries per second (QPS) asyncio: 70 QPS 🚫 Polarway + Tokio: 1200 QPS 🔥 ⏱️ <1ms 🚀 100k+/core 💾 O(batch) ⚙️ Linear CPU 17× faster ⚡
Rust Tokio Async Lakehouse April 5, 2026

Zero-Cost Async: From Python asyncio to Rust's Tokio Runtime

Python's GIL saturates at 70 QPS. Polarway uses Rust's Tokio runtime to reach 1200 QPS with work-stealing across all cores, $O(\text{batch})$ memory, and zero exceptions — only Result<T,E> and Option<T> monads...

⭐ Featured Execution time (seconds) Python (NumPy) 120s Numba JIT 36s Rust + PyO3 0.8s ⚡ 150× faster
Rust PyO3 Performance March 14, 2026

Rust in Python: How We Achieved 150× with PyO3

Python dominates quant finance for its ecosystem, but when the inner loop is a Monte Carlo simulation with $10^6$ paths, interpreter overhead becomes the bottleneck. We solved this with Rust + PyO3: the critical 20% of our code runs in Rust...

P&L = ∫½Γ(σ²−σₕ²)dt Asset price Continuous rehedging → profit without direction Direction = 0
Delta Hedging Rough Volatility Gamma Scalping April 5, 2026

Volatility Arbitrage: A Quantitative Framework for Delta-Neutral Trading

From Black-Scholes to Rough Heston — a mathematical guide to profiting from volatility mispricing without directional risk. The 3 volatilities, delta hedging, gamma scalping, Volterra processes, and empirical validation...

∂V/∂t + ℋ = 0
HJB · Regimes · Execution costs
WP3 HJB Stochastic control April 19, 2026

Adaptive Portfolio Construction & Execution

Regime detection, HJB allocation and execution costs unified in a single stochastic-control recursion. Sharpe +1.42, turnover −38%, drawdown −27% on a synthetic 5-asset example...

m(t) HJB ← → FP Mean-Field Nash Equilibrium
MFG McKean-Vlasov optimiz-rs March 22, 2026

Mean-Field Games & Portfolio Optimization

McKean-Vlasov dynamics, particle methods, optimal quantization and Differential Evolution in Rust — a complete portfolio optimization framework with optimiz-rs and Polarway...

Strike (K) σ_imp H ≈ 0.1
Volatility Options March 12, 2026

Arbitrage Strategies with Rough Volatility

Classic models (Heston, SABR) assume H = 0.5. Empirical data shows H ≈ 0.1. This roughness has direct implications for option pricing and variance arbitrage strategies...

validate transform persist ✓ Ok ✗ Err Railway-Oriented Programming
Rust Functional March 10, 2026

Railway-Oriented Programming with Rust & Polarway

How Rust's Result<T, E> pattern and functional combinators eliminate runtime errors in high-frequency data pipelines. Introduction to Polarway...

Exchange WebSocket gRPC Tonic (320µs P99) LRU Cache Parquet DuckDB Delta Lake (ACID + Time Travel)
gRPC Delta Lake Lakehouse March 20, 2026

Polarway: gRPC Tonic + Delta Lake for HFT

Why gRPC beats REST for HFT (320µs, Protobuf, bidirectional streaming), 4-layer architecture with Tonic + Arrow zero-copy, ACID Delta Lake, time-travel for MiFID II/GDPR, and right-to-erasure with VACUUM...

📄 Go Deeper with Our White Papers

Enjoyed our blog? Our technical papers offer rigorous mathematical analysis with detailed worked numerical examples.

FREE

Mean-Field Games & MCMC

Multi-agent optimal control, Bayesian estimation, Almgren-Chriss execution.

Read Documentation → MFG Tutorial →
FREE

Rough Heston & Signatures

Fractional volatility, SVI calibration, vol arbitrage, path signatures.

Read Article →
FREE

Adaptive Portfolio

Cointegration, HMM, spectral clustering, optimal stopping, risk parity.

Read Article →

📬 Stay Informed

Receive our new research articles directly in your inbox.