Yu (Candice) Song

I’m very pleased to share that our paper, “Microstructure Noise Decomposition: A Liquidity-Adjusted Volatility Estimator,” has been published for publication in Finance Research Letters. I would like to express my sincere gratitude to Seungho Shin for his wonderful collaboration and invaluable contributions throughout this study. In this study, we develop a liquidity-adjusted range based volatility estimator that incorporates bid–ask spread information to better account for microstructure noise and liquidity-induced variation in financial markets. Our paper: https://lnkd.in/ewV8SkQN

43

AI is already inside modern HFT, but not as an LLM making buy/sell decisions. At leading quantitative trading firms, ML and deep learning are increasingly embedded across the trading stack: - Alpha & short-term prediction: extracting signal from order flow, limit-order books and high-frequency market data. - Market making & pricing: estimating fair value, liquidity, fill probability and adverse selection. - Execution: optimizing when, where and how to trade under market-impact and latency constraints. - Risk Management: adapting exposure to volatility, liquidity and changing market regimes. - Quantitative Research: accelerating hypothesis generation, feature discovery, backtesting and model evaluation. - Engineering: AI-assisted research infrastructure, code, data analysis and automation. The interesting part is what happens next. Modern HFT still is Mathematics + Statistics + Market Microstructure + ML + Low-Latency Systems And LLMs? They are far more likely to transform the research and engineering layer than replace the nanosecond trading engine. The real competitive advantage may be the ability to move faster through the entire loop: Data → Research → Model → Backtest → Execution → Feedback → New Research The next generation of HFT researchers may therefore need to understand both the mathematics of markets and the machinery of AI. AI in Quant Finance is no longer a future topic, it is becoming part of the HFT stack. #AI #HFT #QuantFinance #MachineLearning #QuantTrading #AlgorithmicTrading #MarketMicrostructure

5

1 Comment

Excited to announce that our paper “Bayesian nonparametric inference in bank business models with transient and persistent cost inefficiency” has just been published in the Journal of Econometrics! 🎉 This is joint work with Dimitris Korobilis and emmanuel mamatzakis. In this study, we develop a new way to classify how banks adapt their business models over time, while separating temporary from long-term inefficiencies. Applying it to European banks, we uncover four distinct business model clusters offering valuable insights for both performance and policy.

157

10 Comments

We are looking for students who want to solve problems in options that do not have a clean answer. The Institute of Quantitative Finance is opening enrollment for the Fall 2026 Options School, jointly with the Options Project of the Quantitative Analytics Program @ Center of Mathematical Finance. Led by Anton Filatov, Head of Derivatives Trading and Head of Options Projects at CMF since 2016. 🌍 100% online | Free | Weekends | September–December 2026 | English START WITH AN UNCOMFORTABLE QUESTION Suppose your options backtest shows a Sharpe of 3. Why should anyone believe it? WE BUILD THE WHOLE SYSTEM Not just a pricing model. Not just a strategy. Not just a backtest. The infrastructure connecting them. MARKET → DATA → SURFACE → STRATEGY → EXECUTION → HEDGE → RISK → P&L And every link has to survive scrutiny. Exchange & Backtesting Build the order book, matcher and wire protocol. The backtester should be something you can inspect: not a black box producing an equity curve. Volatility Curve Engine Build curves from sparse strikes, wide spreads, stale quotes and missing sides. Strategy Engine A strategy becomes a formal specification: signal → package → sizing → vega/gamma → limits → execution → costs → P&L. Liquidation & Vega Hedging Estimate what it costs to exit. Model own-flow impact. Measure liquidity as vega capacity by bucket. Understand what remains after the hedge. Crypto Options Take the framework into a market with: 24/7 trading, inverse coin-margined contracts, perpetual funding, and option-implied carry. AI Agent Layer Build agents directly on top of the quantitative infrastructure. If an agent cannot be inspected and replayed command by command, it is not part of the system. THIS IS FOR PEOPLE WHO LIKE THE HARD PART We are interested in people coming from: Mathematics · Statistics · Computer Science · Physics · Economics · Quantitative Finance · Machine Learning · Software Engineering · Trading · Research THE PEER GROUP MATTERS The strongest part of a quantitative program is rarely the syllabus: it is the people in the room. We would rather have 20 exceptional people than 200 passive participants. START HERE Intro Lectures Advanced Backtesting Engine for Options Trading https://lnkd.in/dB2iunQD Quantitative Options Trading: The Beginning https://lnkd.in/d8erq8aa QA Session https://lnkd.in/dhP82mBf Advanced Options Trading: Gamma https://lnkd.in/d-crczFh Options Greeks Overview https://lnkd.in/dFChzBdt xVA https://lnkd.in/dsWXthts Intro Literature Antonio Castagna — FX Options and Smile Risk, Chapters 1.1–1.2 https://lnkd.in/d5qgGn82 Jiles Jewitt — FX Derivatives Trader School, Chapter 1 https://lnkd.in/daf6NbYj Richard Flavell — Swaps and Other Derivatives, pp. 1–30 https://lnkd.in/dMxMfYFY APPLICATION CV + Motivation Letter → Project Challenge → Live Interview Selected applicants receive the project challenge by email. 📅 Registration closes September 25, 2026 https://lnkd.in/d4w46nZY

26

1 Comment

I’ve been exploring a path-dependent approach to credit risk modelling that moves beyond standard Markovian intensity frameworks. Using the Signature transform from Rough Path Theory, the model encodes the full history of credit spreads and market factors, allowing non-linear dynamics to be handled through linear models in signature space. The framework naturally captures rough volatility effects and sequence-dependent stress patterns, with direct relevance for IFRS 9 Point-in-Time calibration and SICR detection. #CreditRisk #QuantitativeFinance #RoughPaths #IFRS9 #RiskModelling #AppliedMathematics

21

UCL Mathematician Riaz Ahmad to lead a shared graduate course with the The Fields Institute For Research In Mathematical Sciences, titled "Advanced Topics in Mathematical and Computational Finance". https://lnkd.in/eRqzEg9R

27

1 Comment

The post highlights how Maximum Likelihood Estimation (MLE) is used to calibrate stochastic interest rate models, a key component of modern fixed-income and quantitative finance frameworks. For a step-by-step learning approach and hands-on understanding, explore the Financial Engineering programs at Quantie: www.quantie.in

46