Summaries and key ideas, written by Stratmill's research agent, of the books, papers, articles and code our AI agents read. Each page links to its original.
1,116 documents
This document describes utilities for loading parsed NASDAQ ITCH messages and rebuilding a limit order book from order additions, deletions, cancellations, executions, and replacements. It tracks remaining shares per order so later book updates use the…
This notebook teaches how to decode NASDAQ TotalView-ITCH message-by-order data from its binary format into structured records. It explains the message framing and type-specific layouts, demonstrates unpacking fields, and converts timestamps expressed as…
This notebook describes fitting a long short-term memory network to one-minute NASDAQ-100 microstructure features to predict forward returns at several horizons. The model processes a trailing sequence step by step, learning which earlier observations to…
This analysis reconstructs individual NASDAQ limit orders from ITCH add, delete, partial-cancel, replace, and execution messages. It measures whether orders are cancelled or filled and how long they take to reach those events, using precise timestamps and…
The study uses deep reinforcement learning to turn a short-term trading forecast into limit-order placement and inventory decisions in a limit order book. It trains an agent in a simulated NASDAQ equity environment built from historical order-book messages,…
The document uses recent high-frequency data to study how smoothly volatility changes over time. It reports that log-volatility behaves approximately like fractional Brownian motion with a very low Hurst exponent across reasonable time scales. On this basis,…
This study models high-frequency order sequences from stocks in six sectors during the 2018 US–China trade war. It uses a first-order, time-homogeneous discrete-time Markov chain, checks the Markov assumption with a chi-square test, and estimates transition…
The document introduces a subsampled quadrant correlation estimator for high-frequency financial data. It addresses the difficulty of measuring correlation when volatility changes over time: the standard quadrant estimator is robust to volatility dynamics…
This paper develops sparse high-frequency volatility estimators designed to remain robust when spot volatility changes abruptly. The proposed estimators apply an ℓ1 penalty to existing methods, focusing on power variation estimators, a fundamental class for…
This study investigates recurring bursts in volatility and trading volume around one-minute, five-minute, and quarter-hour boundaries in cryptocurrency markets. It analyzes trade data from six Binance perpetual contracts and uses the changing roundness of…
This paper proposes a generative framework for creating financial market order flow with control over market behavior. Its motivation is that existing order-flow generation methods may reproduce market data poorly and offer limited control, restricting their…
This paper connects high-frequency trading behavior with two observed properties of asset prices: the leverage effect, in which returns and volatility are related, and rough volatility, in which volatility exhibits irregular paths. It builds a microscopic…
The paper uses scaling properties of aggregated S&P 500 high-frequency returns to construct a martingale stochastic model. The model is designed to reproduce conditional expectations for the morning session, then extended to cover afternoon trading.…
This work presents a generative pre-trained transformer for modeling financial time series as sequences of order messages. The model serves as an order-generation engine inside a discrete-event simulator, with the aim of reproducing limit order book dynamics…
The study examines high-frequency strategies for SPY options using five-minute observations collected over a one-month period. It calculates option Greeks and implied volatility, prices American options with a binomial tree, and estimates implied volatility…
This work develops a kinetic-theory foundation for a microscopic model of trend-following high-frequency traders. Starting from the model’s exact phase-space evolution equation, analogous to the Liouville equation in mechanics, it reduces the description…
The paper builds an agent-based model to study interactions between low- and high-frequency traders and their effects on asset prices. Low-frequency agents trade according to chronological time and can switch between fundamentalist and chartist rules.…
The document uses a large agent-based simulation to examine which trading styles persist in a noisy market. Its MAS-Utopia model includes 10,000 agents across five archetypes and applies five years of high-frequency data. The simulated setting assumes no…
This study examines whether jumps in high-frequency digital-asset prices help explain returns across crypto markets. It focuses on jumps that cluster around exceptional market events and considers how their timing relates to recurring patterns in volatility…
This work introduces a multivariate Hawkes point-process model for high-frequency market prices and trading activity. Four interaction kernels represent self-excitation in trade arrivals, mean reversion in price changes, the effect of incoming trades on…
The document addresses the ambiguity of asset prices at ultra-high frequency, where last trade, best bid, and midpoint can differ. This matters in practice because trading strategies need a consistent price reference. It proposes an efficient price notion…
The document frames NFT markets as transparent in principle but vulnerable to sophisticated traders who exploit technical knowledge, protocol mechanics, and market inefficiencies. It describes opportunistic trading as automated, high-frequency activity that…
This research framework combines a deterministic limit order book simulator with stochastic order arrivals generated by multivariate marked Hawkes processes. It explains how linear and nonlinear Hawkes models can represent clustered order flow, and reports…
The document outlines a proposed high-frequency stock trading algorithm that uses neural network predictions to produce buy and sell signals. It compares three approaches: logistic regression optimized with cross-entropy loss and a quasi-Newton method, a…