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.
1,033 documents
This document describes a continuous-time model of a single asset traded by heterogeneous agents. Its second-order dynamics represent delays between investment decisions and trades. The model includes value-based mean reversion, speculative or contrarian…
The study tests the weak form of the efficient market hypothesis for Bitcoin with a feedforward neural network. It asks whether adding returns or related features from other assets improves prediction, alongside Bitcoin returns. The additional inputs include…
The document introduces price impact, explains how it is measured and modelled in recent research, and questions the intuitive idea that trades move prices through a simple, direct mechanism. It highlights empirical evidence that signed order flow and price…
This study uses the Hurst exponent to classify a time series as trending or mean reverting, then selects a strategy intended to suit that behavior. Momentum is matched to trending conditions, while mean reversion is used when prices exhibit mean-reverting…
The study compares 41 machine learning models, split between classifiers and regressors, for predicting Bitcoin prices and assessing their use in algorithmic trading. It combines prediction measures such as absolute and squared error with trading measures…
This work develops a trading strategy for an FX broker liquidating a large position in an illiquid currency pair. The broker trades a triplet comprising that pair and two liquid pairs, selected so that one pair is redundant. The setup is risk-neutral but…
This study formulates speculative trading as a sequential optimal stopping problem, choosing entry and exit times under general utility functions and price processes. It first relaxes the stopping problem by representing entry and exit events as jumps of Cox…
This document presents explicit solutions for a class of linear backward stochastic differential equations driven by Gaussian Volterra processes. The processes discussed include multifractional Brownian motion and multifractional Ornstein-Uhlenbeck…
This discussion explains why dollar-neutral quantitative strategies, including statistical arbitrage in equities, experienced substantial drawdowns during the COVID-19 market selloff. It outlines why such strategies can work in ordinary conditions,…
This paper studies whether independent deep reinforcement learning agents in an optimal-execution game develop behavior consistent with collusion. In a two-player, finite-horizon liquidation setting, agents use proximal policy optimization and have access to…
The document presents an exact solution for a simple trend-following strategy and describes the resulting distribution of profit per trade. Rather than treating outcomes as symmetric, the model produces an asymmetric, option-like distribution whose shape…
This paper introduces compound and regime-switching compound Hawkes processes as models for price dynamics in limit order books. Hawkes processes represent event arrivals whose rates can depend on prior events; the compound variants are used here to connect…
This document describes an extension of the minority game in which the market rewards contrarian strategies when it is far from efficiency and trend-following strategies when it is close to efficiency. The model links the relative success of each trading…
This paper makes the information used in mean-variance portfolio choice part of the optimization problem. It argues that standard portfolio analysis can be distorted by including variables unavailable at decision time or by mistaking shared variation for…
This paper develops a transaction-monitoring framework for detecting laundering in Rwandan mobile money. It addresses severe class imbalance, delayed and limited labels, and finite investigative capacity. Using a synthetic dataset of more than nine million…
This study asks whether cryptocurrency investors trade differently from stock investors and uses asset price histories as evidence of those behaviors. It first applies multiple machine-learning models to classify cryptocurrency and stock time series observed…
This paper studies two investors who choose mean-variance portfolios while competing over relative wealth. One investor knows the true stock dynamics; the other must infer them from observed market evolution. Their decisions are linked because each investor…
This research develops a pairs-trading approach for cointegrated cryptocurrency pairs using copulas to model dependence and identify relative mispricing. Pair selection combines linear and nonlinear cointegration tests with a correlation measure. The…
This study uses an event-study approach to examine how SEC announcements classifying crypto assets as securities affect the returns and trading volumes of explicitly named assets. It reports substantial negative returns after announcements, with the decline…
This study examines a low-minus-high rank effect in commodities: lower-priced, lower-ranked commodities have higher subsequent returns than higher-ranked ones. Using nonparametric econometric methods, it argues that this effect follows from a stationary…
This study examines whether daily Twitter sentiment can help predict stock returns and support trading decisions. It combines machine learning to assess the sentiment signal with reinforcement learning, specifically Q-learning, to derive a trading policy…
The document describes an equities strategy that it says can generate outsized returns for a sufficiently large, market-neutral quantitative hedge fund. It also frames the strategy as contributing to greater global wealth inequality, making the…
MiniFed is a framework for simulating Federal Open Market Committee meetings with large language model agents. It aims to represent meeting members and the committee’s decision process, shifting attention from studying only the effects of Federal Funds rate…
The paper develops stochastic-volatility models that use opening and closing prices together with each period’s observed high and low. The motivation is that extreme prices within a period contain information about volatility that close-to-close observations…