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.
5,018 documents
The document surveys a quantitative research platform’s components and workflow. It describes tools for strategy development, financial data access, factor research, model training, backtesting, simulated trading, and live execution. Its examples span stock,…
The document summarizes research on applying conditional generative adversarial networks to calibrate individual systematic trading strategies and combine them into ensembles. The proposed workflow covers training and selecting a cGAN on time-series data,…
The article questions the use of deep learning to forecast stock or futures returns from historical indicators. Its author says they tried both classification and regression approaches, including recurrent networks, and found the results poor. They also…
This article explains attention and Transformer components, then adapts them to predict stock returns from factor time series. Its proposed model adds Time2Vec features to factor inputs, applies self-attention and multi-head attention, and stacks encoder…
The document describes Lumibot as a Python framework for creating rule-based strategies, AI-assisted trading systems, and hybrid approaches. Conventional Python logic can handle indicators, schedules, position sizing, and risk controls, while AI agents can…
This reference surveys machine-learning methods and relates them to investment research tasks. It covers linear and logistic regression, naive Bayes, nearest neighbors, support vector machines, decision trees and ensembles, neural networks, sequence models,…
This introduction presents FinRL as an open source framework for applying deep reinforcement learning to financial trading. Its stated design aims include modular components that can accommodate different markets and data sources, configurable rather than…
The code implements a time-series prediction model paired with a TRA component that can combine predictions across multiple states. A base model, such as an LSTM, produces hidden representations and state-specific predictions. When multiple states are…
This report summary examines how to select timing variables and forecast equity factor returns within a multi-factor strategy. It proposes Lasso and Elastic Net regression for screening predictors and estimating factor returns. It also describes extensions…
This legacy platform tutorial outlines a workflow for predicting stock direction with a hybrid convolutional and recurrent neural network. It labels each observation according to whether the following ten-day return is positive or negative, builds rolling…
This code collection provides helper functions commonly used when building quantitative factors from time-series data. It wraps TA-Lib operations for rolling sums, standard deviations, minima, maxima, momentum differences, and weighted moving averages. It…
The study applies machine learning to estimate an optimal rebalancing frequency for pairs trading. It uses minute-level prices for 50 large cryptocurrencies from Binance during 2022 and 2023. For each asset pair, a simulated pairs-trading algorithm…
This code implements a bivariate Gumbel copula for representing dependence between two uniform variables. It provides methods to generate paired samples from independent uniform inputs, calculate the copula density and cumulative distribution, and evaluate a…
This Chinese-language article curates research papers and books on reinforcement learning (RL) in finance. Its examples span trading and market applications, Q-learning combined with Black–Scholes, option hedging and dynamic replication, portfolio…
This tutorial explains term frequency–inverse document frequency (TF-IDF), a method for assigning greater weight to words that occur often in one document but less often across a corpus. Term frequency captures within-document prevalence, while inverse…
This study explains combinatorially symmetric cross-validation (CSCV) as a way to estimate the probability that a strategy selected as best in a backtest will rank poorly across alternative partitions of that history. The backtest period is divided into…
The article explains naive Bayes classification and applies it to a stock selection strategy. The classifier estimates the probability of each outcome from historical labeled observations, assuming features are conditionally independent and equally…
This brief Chinese-language forum entry asks whether stock-selection backtests using random forest classification and a StockRanker ranking model can be trusted when features were not standardized or treated for outliers. The accompanying tag states that…
This configuration describes a Qlib workflow that trains a PyTorch deep neural network on Alpha158 features for CSI 300 stocks, using the Shanghai 300 index as its benchmark. It defines training, validation, and test periods, drops the VWAP feature, fills…
The document outlines an example trading system that uses two support vector machines, one for buy signals and one for sell signals. It initializes seven technical indicators, passes their handles as model inputs, and derives training inputs and target…
This configuration specifies a Qlib workflow for training an ADARNN model on CSI 300 equities in the Chinese market. It uses Alpha360 features, robust z-score normalization with outlier clipping, feature filling for missing values, and cross-sectional rank…
The document describes how machine learning is being used by institutional trading teams for execution, risk management, and signal research, with digital assets as a key setting. It argues that model quality depends on access to reliable, detailed data from…
This article examines how training-window length affects AI stock-selection models, comparing multi-year datasets with shorter windows and proposing rolling training and evaluation on a designated validation period. It reports that, in its experiments,…
The document presents a Chinese-stock ranking workflow embedded in a report about a Papermill execution error that arose when submitting a strategy for simulated trading. It constructs price, volume, valuation, and technical features, creates a…