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 addresses portfolio construction for statistical arbitrage by seeking portfolios whose returns or spreads exhibit strong mean reversion. The optimization balances a criterion for mean-reversion strength with portfolio variance and an investment…
This document describes a stochastic control framework for choosing trades in commodity futures over a finite investment horizon. It builds on the Schwartz 1997 model, which represents commodity prices using a stochastic convenience yield, and formulates…
This note studies utility maximization when trading incurs proportional transaction costs. Its central aim is to prove a limit theorem for optimal trading strategies in that setting, addressing an open question identified by the authors. The result concerns…
The document introduces TrendFolios, a portfolio construction framework that combines momentum and trend-following signals across several asset classes and asset-class risk factors. It describes an active investment strategy and says the framework is…
This paper presents a model-free reinforcement learning framework for reallocating a portfolio among financial assets. Its components include an ensemble of independent evaluators, memory of prior portfolio weights, online stochastic batch learning, and an…
This document studies how to manage positions across several mean-reverting assets, including whether an asset with no mean reversion can still belong in the portfolio. It formulates position management as an optimal control problem for an investor with…
This theoretical study asks when several causal risk mandates, each built from overlapping information, can be implemented by one self-financing portfolio that is optimal given pooled information. In an incomplete continuous Brownian market, it projects…
This document outlines a portfolio management framework that pairs a return-seeking reinforcement learning agent with a barrier-function risk controller. The learning component searches for profitable investment decisions, while the controller monitors…
This study formulates pairs trading as a portfolio optimization problem for two cointegrated stock log-prices. A singular stochastic control model jointly determines when to trade and how many shares to adjust, while incorporating proportional transaction…
This article studies optimal investment and consumption in an incomplete market when preferences are expressed through nonlinear g-expectations. Trading strategies are subject to general constraints, so the investor’s choices must satisfy restrictions as…
The document describes constructing strategies by classifying market states instead of directly forecasting prices or returns. Labels define states, and market features such as moving averages are selected for their ability to distinguish those labels. A…
The document examines whether models that map market conditions directly to portfolio weights can outperform conventional cross-asset futures strategies. It describes training these policies on sixteen liquid CME futures with a differentiable Sharpe-ratio…
The paper presents a general framework for deriving an optimal trading strategy from a historical indicator. It constructs one strategy intended to maximize expected return and a separate strategy intended to maximize the information ratio. The…
The document studies semi-static trading strategies that combine dynamic trading in a liquid asset with static, buy-and-hold positions in options on that asset. It focuses on the mathematical properties of the set of outcomes such strategies can produce when…
This document formulates expected utility maximization of terminal wealth for an investor trading futures in discrete time. It places the problem in a Brownian market framework and introduces practical futures concepts including margin, gearing, and…
This framework links a target-date fund’s investment objective to pension design inputs, such as contribution rates, working years, retirement age, and replacement-rate goals. Instead of prescribing age-based asset limits alone, a sponsor specifies a…
The paper questions whether combining trend signals across many horizons always improves diversification. It dynamically selects horizon weights for each asset using Bayesian optimization, then applies sparse Bayesian graphical models with turnover control…
The document describes an equity factor strategy that tilts a global stock portfolio toward companies whose ESG ratings have risen and away from those whose ratings have fallen. ESG momentum is measured over the prior 12 months, and the portfolio is…
The document describes a monthly strategy that blends momentum across equity factors with a broad market portfolio. It forms fast and slow signals from each factor’s recent one-month and twelve-month returns, ranks signal magnitudes to allocate factor…
The document describes a cross-sectional commodity futures strategy based on return asymmetry. It defines an IE measure as the difference between the counts of unusually large positive and negative daily returns, using a rolling 260-day window. At each month…
The document describes a U.S. stock market strategy that sorts companies by market capitalization and then ranks them by return on assets (ROA). ROA is calculated from quarterly income and assets from the prior quarter. The portfolio buys the three highest…
The document explains a currency carry trade: borrow or short currencies with relatively low central bank rates and hold currencies with relatively high rates, aiming to earn the interest-rate differential. Its simple example forms a universe of 10–20…
This strategy identifies equity industries whose estimated alpha relative to the broad market has become statistically significant, interpreting the break as a possible bubble. Using roughly a decade of historical returns, an investor estimates alpha with a…
This strategy ranks five ETFs representing US stocks, foreign stocks, bonds, real estate, and commodities by their trailing 12-month returns. It selects the three strongest, weights them equally, holds them for one month, then repeats the ranking and…