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.
765 documents
The document presents KBPT, a pairs trading policy that combines a Kalman filter, Bollinger Band signals, and a neural network based on the KalmanNet architecture. Traditional versions model the relationship between two assets as a linear state-space process…
This work examines ways to reduce latency in performance-sensitive C++ applications, with high-frequency trading as its main setting. It describes a programming repository with benchmarked examples, an optimized market-neutral statistical arbitrage pairs…
This document describes a model-agnostic framework for forecasting asset volatility by extracting a compact set of time-varying factors from realized volatilities. The factors are intended to capture changing cross-asset co-movement while keeping computation…
This study develops a Monte Carlo method for pairs trading when the spread follows a mean-reverting Ornstein–Uhlenbeck process driven by jumps. It uses a variance gamma process, which allows infinite activity and more flexible spread dynamics than the…
This article examines correlations and co-movements among stocks to study intraday seasonality and market evolution. Using intraday CAC 40 data, it revisits earlier findings that average stock correlations rise over the course of the trading day. It applies…
This paper studies when to close a pair trade whose spread is modeled as an Ornstein–Uhlenbeck-type process driven by a finite-activity Lévy process. The jump component extends a setting where the asset difference is modeled without jumps, allowing the…
The paper proposes a futures pair-trading strategy that uses path signatures as time-series features. It decomposes the signature into two indicators: a segmented signature measure of path interactivity and a covariation-of-increments measure of direction.…
This document describes a deep-learning framework for identifying time-varying relationships between assets for pair trading. Multi-modal Temporal Relation Graph Learning combines time-series observations with discrete features in a temporal graph. A…
The study adapts a pairs trading framework to Polish equities by comparing an asset with a replication built from risk factors, rather than pairing it with a second highly correlated asset. It constructs factors using Principal Components Analysis, exchange…
The document presents a portfolio construction method for pairs trading that limits overlap between selected pairs. It represents assets as graph nodes and cointegrated candidate pairs as weighted edges, with edge weights reflecting the strength of…
The study examines the statistical basis of Bollinger Bands by connecting them to a rolling regression time-series model. It applies the framework to pairs trading and derives a relationship between trade duration and returns for strategies based on the…
This preliminary analysis applies an Ornstein–Uhlenbeck process to model spreads between paired stocks for a mean-reversion trading strategy. It compares that approach with a simpler method that estimates the spread’s mean and standard deviation over a…
This paper develops a state-space approach to modeling mean-reverting spreads used in statistical arbitrage, including pairs trading. It treats observed spreads as noisy readings of hidden states and uses real-time estimates of those states to identify…
This study tests a pairs strategy on spot prices for 17 Indian commodities across energy, metals, and agriculture from 2010 through 2018. It uses Johansen cointegration tests on the training sample to select long-run related commodity pairs, then models each…
The paper addresses a portfolio-construction problem in statistical arbitrage: signals from many pairs can conflict, making it difficult to combine opportunities across a large group of securities. It proposes representing preference relations among…
The document studies portfolio growth optimization when expected asset returns are uncertain. It considers an incomplete, high-dimensional market in which asset prices depend on stochastic factors. Rather than fixing the drift completely, the framework…
The paper develops an adaptive matrix-variate Markov chain Monte Carlo method for Bayesian cointegrated vector autoregressions. It replaces griddy Gibbs sampling with an automated Adaptive Metropolis approach, intended to make posterior estimation practical…
The paper studies estimation of mean-reverting price dynamics from high-frequency observations, where market microstructure noise can bias parameter estimates. It proposes a maximum-likelihood estimator for an Ornstein–Uhlenbeck process that accounts for…
This study develops a timing method for trading a mean-reverting price spread, motivated by pairs trading. It models the spread as an Ornstein–Uhlenbeck process and frames opening and closing a position as a double stopping problem, with transaction costs…
These notes build a quantitative treatment of mean-reversion, beginning with pair trading and then progressing through demeaning, regression, weighted regression, constrained optimization, and factor models. The sequence shows how increasingly general…
The study examines liquid commodity futures traded in Chinese markets before and after the onset of Covid-19. It compares empirical return moments and cointegration across products, motivated in part by the relevance of cross-commodity relationships to pairs…
This work develops a Kalman filtering method for settings where noise covariance matrices and predicted covariance estimates are uncertain. It uses a distributionally robust formulation to account for plausible alternative models, with bicausal optimal…
This study presents a model-free reinforcement-learning framework for statistical arbitrage between paired stocks. It first constructs a mean-reverting spread by choosing asset coefficients that minimize an empirical measure of how long the spread takes to…
This document describes CREDIT, a reinforcement learning approach to pair trading that uses a bidirectional gated recurrent unit and temporal attention. The architecture is intended to capture relationships across time in the price movements of two assets,…