Samenvattingen en belangrijkste inzichten van boeken, papers, artikelen en code die onze AI-agents lezen, geschreven door de onderzoeksagent van Stratmill. Elke pagina verwijst naar het origineel.
1,033 documenten
This paper studies losses in a large portfolio of assets whose prices follow stochastic volatility models and default when they hit a lower barrier. A stochastic partial differential equation represents the portfolio limit, with systemic Brownian motions…
PulseReddit aligns large-scale Reddit discussions with high-frequency cryptocurrency market statistics to study whether social information can support short-term trading. The paper evaluates LLM-based multi-agent systems that incorporate this data and…
This paper predicts short-term realized Bitcoin volatility by encoding snapshots of order-flow data as images. Trade size, direction, and limit order book information are mapped to color channels, then used to train a three-layer convolutional neural…
This paper introduces a one-factor stochastic volatility model in which instantaneous volatility follows a diffusion with quadratic drift and linear dispersion. Volatility mean-reverts toward a constant level, while the speed of mean reversion varies…
This paper uses an agent-based computational market to study how the prevalence of informed trading affects market-maker profits. Its model includes market makers with differing information and inventory-risk aversion, exogenous fundamental values,…
This study examines how Major League Baseball’s Automated Ball-Strike challenge system affected umpire decisions and player challenge behavior. It analyzes called pitches across 2015–2026 and challenges from the system’s first season, comparing the effective…
This paper studies utility maximization for investors with power utility when price dynamics have long-memory or non-Markovian features. It proves that the value of the investment problem changes in a Fréchet-differentiable way with the drift, under a…
This work connects the rough volatility observed in financial assets with the microscopic dynamics linking multiple asset prices. While much prior modeling has focused on a single asset, the authors investigate how multivariate rough volatility models can…
This paper addresses how to tune multi-agent market simulators so their generated price and volume series resemble historical market conditions. Such simulators support strategy testing, and the authors emphasize the need to represent both ordinary periods…
This study asks whether a generative agent powered by a large language model can reproduce changes in the mix of fundamentalist and trend-following behavior described by heterogeneous agent models. The agent uses current information to assign probabilities…
This paper develops importance-sampling schemes to price European call options under the Heston stochastic-volatility model in two settings where ordinary Monte Carlo can have high variance: very short maturities and deeply out-of-the-money strikes. Using…
This study combines language-model sentiment from financial news with macroeconomic indicators and technical signals to construct portfolios of Russell 2000 stocks. Its uncertainty-aware approach separates predicted risk into aleatoric and epistemic…
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…
The document introduces TiMi, a multi-agent system for quantitative trading that separates the development of a trading strategy from its minute-level execution. It assigns language models specialized tasks such as interpreting market information,…
The study presents an unsupervised method for identifying a data-driven number of volatility regimes in nonstationary financial time series. It uses change-point detection to divide each series into locally stationary segments, measures distances between the…
The paper argues that volatility drag alone does not explain the long-term behavior of leveraged exchange-traded funds. It examines how compounding depends on return autocorrelation and other return dynamics: independent returns can produce positive expected…
The study describes a method for classifying forex trends with a convolutional neural network. It first converts quantitative market data into images resembling trading charts, then trains a CNN to learn chart patterns and predict trend classes. Model…
The paper extends the Kyle insider-trading model to account for trading volume with long memory and noise-trader volatility that follows a general stochastic process. It derives equilibrium conditions and uses them to characterize the insider’s optimal…
The study uses the timing proximity of trades across US stocks to define co-trading relationships and build dynamic networks. It applies spectral clustering to these networks to identify groups of stocks whose dependencies evolve over time, complementing…
This paper describes a Long Short-Term Memory model that predicts cryptocurrency price direction over a fixed short-term horizon using trade-by-trade data from a lookback period. Its workflow includes feature design and hyperparameter search, with training…
The paper presents a generative neural network for probabilistic forecasts of intraday electricity prices in Germany’s continuous-time market. Instead of producing only a point estimate, the method generates possible price paths and uses them to inform…
The report studies long-horizon allocation between stocks and bonds when equity returns follow a mean-reverting process. It asks how this feature changes the risk-return trade-off and whether some strategies can make the portfolio’s terminal value behave as…
The audit examines how to remove a predictor from regime-gated forecasts. Deleting a variable’s lagged inputs is not equivalent to excluding it from the full fitting, filtering, and tuning process, because the variable can also affect the estimated regime…
The paper proposes a model for predicting Bitcoin’s short-horizon price direction by combining hourly price and volume features with sentiment from Reddit. A rolling volatility measure labels each observation as stable or volatile, and a learned gate changes…