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.
5,922 documents
This paper presents Multi-Factor Inception Networks (MFIN), a framework for trading multiple cryptocurrencies using price returns and other frequently updated signals. It motivates adding factors such as network hashrate and search interest to address the…
The study explores deep reinforcement learning as a way to develop an adaptive trading strategy for a portfolio of 30 stocks. It uses daily prices as the training data and trading environment, then compares the agent with the Dow Jones Industrial Average and…
The paper examines whether coherent risk measures can control investors who have limited liability or seek excessive tail risk. It introduces risk-measure-specific statistical arbitrage, termed rho-arbitrage, and argues that when such opportunities exist,…
The study defines price momentum using analogies to physical velocity and mass, then uses this measure to select assets for weekly contrarian strategies. It applies the approach to South Korea’s KOSPI 200 and the US S&P 500 universes, comparing the resulting…
The study builds a changing cryptocurrency network using two kinds of links: return cross-predictability and technological similarity. It applies a dynamic covariate-assisted spectral clustering method to estimate communities while accounting for both…
The paper proposes a dynamic cryptocurrency portfolio method that combines technical signals with news sentiment. It uses the 14-day Relative Strength Index and Simple Moving Average to represent momentum, while VADER analyzes news sentiment and Google…
The paper adapts statistical arbitrage to real-time display advertising, where impressions with similar expected effectiveness can carry different prices across fragmented marketplaces or pricing arrangements. Its method links cost-per-action campaigns with…
The document describes a systematic trend-following strategy for commodity futures that combines conventional single-market trend indicators with cross-market signals. These cross-sectional signals aim to capture momentum spillover, where trends in one…
The document proposes mapping real exchange rate movements onto a string-like structure to build forecasting models for forex trading. It argues that conventional long-horizon econometric approaches can miss practical features such as transaction costs,…
This paper describes a follow-up study on forecasting stock option trends with machine learning and applying those forecasts in trading. It examines recurrent neural networks and long short-term memory networks, then evaluates combining multiple models to…
The paper presents a method for operating and valuing electricity storage when intraday prices are uncertain. It formulates the charging and discharging problem as convex stochastic optimization and uses Stochastic Dual Dynamic Programming to solve it. The…
The paper studies whether strategies built on non-equity assets can diversify algorithmic investment strategies tied to the S&P 500. It compares signals generated from LSTM forecasts, ARIMA-GARCH forecasts, momentum, and contrarian rules. The assets…
This article applies randomized controls to empirical asset pricing and investment performance evaluation. It uses geometric random walks, described as a class of Markov chain Monte Carlo methods, to generate random portfolios that satisfy investor…
The paper addresses the asset-ranking step used in cross-sectional systematic portfolios. It argues that simple heuristics and rankings formed by sorting predictions from conventional regression or classification models can be suboptimal, drawing an analogy…
The document outlines a systematic approach to adapting trading strategies and asset allocation as market conditions change. It proposes using state-switching Markov autoregressive models to identify and predict regimes loosely associated with Wyckoff…
The document studies how the indirect utility process associated with a trading strategy changes when the risky asset’s return dynamics are perturbed. The strategy need not be optimal, so the analysis also applies to utility processes arising from suboptimal…
This document presents a Bayesian method for estimating correlation matrices that change over time. It represents dependence with a low-rank factor model, uses latent states with dynamic shrinkage priors for locally adaptive regularization, and models…
This work proposes RADAR, a retrieval-augmented diffusion framework for learning market state representations used in portfolio optimization under the stochastic discount factor framework. Its motivation is that markets shift between regimes, financial…
Tail-GAN is a data-driven scenario generator designed to simulate realistic joint price dynamics for high-dimensional multi-asset portfolios, with emphasis on losses in the tail of the distribution. It uses the joint elicitability of Value-at-Risk and…
The paper applies supervised machine learning to post-earnings-announcement drift (PEAD), an equity anomaly in which returns continue to move after earnings news. Its model uses XGBoost and engineered fundamental and technical features built from quarterly…
This paper extends functionally generated portfolios, a framework within continuous-time stochastic portfolio theory. In the standard formulation, a generating function describes a portfolio’s return relative to a passive benchmark through a pathwise master…
The paper presents a risk-constrained portfolio optimization approach for virtual bidding in U.S. wholesale electricity markets. It forecasts inter-hour locational marginal price spreads with a recurrent neural network, using dependencies created by market…
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 paper develops a way to test whether adding securities or relaxing investment constraints expands the opportunities available to investors who evaluate outcomes using prospect preferences. It formulates a test for prospect spanning between nested…