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
This overview distinguishes language generation from financial prediction, arguing that strong text generation does not by itself solve markets’ low signal-to-noise ratios, changing patterns, and limited data. It explains GPT’s development and instruction…
This documentation page catalogs Freqtrade features and settings that have been deprecated or removed, then explains migration implications for strategies and stored market data. It covers command-line options, pairlist configuration, strategy interfaces,…
This strategy note describes a China A-share approach that blends model-based stock ranking with technical timing. It begins with a manually defined pool of roughly 100–300 large-cap names associated with the CSI 150 universe, then ranks candidates using…
This research summary compares large language models with a financial word-list method for classifying news sentiment and assessing stock returns. It describes analysis of 965,375 US financial news articles matched with daily stock returns. Models classify…
This study builds a text-derived earnings surprise factor, SUE.txt, to capture post-earnings announcement drift in Chinese equities. It uses analyst report titles and summaries related to earnings forecasts, converts them into word-frequency features, and…
The article explains hidden Markov models as a way to infer unobserved market regimes from observable data. It illustrates the idea with Chinese equities, treating market conditions such as bull, bear, and sideways phases as hidden states and returns and…
This educational article introduces operations on matrices and vectors as groundwork for understanding deep learning. It defines elementwise matrix addition, adding a scalar to each matrix element, and broadcasting a vector across matrix rows. It then…
This article adapts a supervised stock-timing approach to Chinese equities. It describes using nine price- and volume-related features to train a binary classifier: the label is positive when the price 30 trading days later is at least the current price, and…
This account of a 2021 industry speech discusses how data, algorithms, and execution shape quantitative investing. It frames potential returns as depending on both market inefficiency and statistical methods that identify patterns robust enough to validate.…
This study evaluates a convolutional neural network that predicts five-day returns across the Chinese stock market. Its baseline uses 2012–2017 data for training and tests on 2018 through October 2021. The authors vary batch size, optimizer, loss function,…
The report combines wavelet denoising with a support vector machine to forecast the CSI 300 index time series. It frames market direction as a classification task and price estimation as a regression task, then uses the forecasts to build both long-only and…
The article examines why A-share strategy results in historical tests can differ from live or simulated trading. It groups causes into data problems such as survivorship bias and price adjustments, model problems such as overfitting and crowded factors,…
This stock-selection method focuses on the metaverse sector and looks for a newly formed KDJ golden cross alongside convergence among five moving averages. It describes screening before 10 a.m. for stocks that are trading that day. The accompanying examples…
This research report examines conditional generative adversarial networks (cGANs) for forecasting asset returns and covariance matrices, then using those forecasts in portfolio allocation. The model conditions on recent multi-asset returns and random inputs…
This document describes a trading system built around reinforcement-learning agents implemented in an MQL library. An agent accepts a vector of predictor values and contains an ensemble of trees; the example supplies normalized closing prices as inputs and…
The document presents an outline for a VeighNa training course on machine-learning-based cross-sectional multi-factor strategies, also described as alpha strategies. It frames these methods as a form of statistical arbitrage applicable to equity selection,…
This Qlib configuration defines a daily stock-ranking experiment for the CSI 300 universe, using the CSI 300 index as its benchmark. Its features combine price and volume transformations, including residual and fit measures, intraday range, and rolling…
This troubleshooting note describes a simulation failure in a single-stock AI ranking strategy: the pipeline reports that no data remains after missing values are removed, even though the strategy completes a backtest. It suggests checking whether the data…
This research review describes a dynamic, out-of-sample process for selecting US active equity mutual funds from observable fund characteristics. It compares elastic net, random forests, gradient boosting, and linear regression. Each year, models are trained…
The document introduces neural networks as a machine learning approach for quantitative stock selection and says they are applied to predicting whether stock prices will rise or fall. It briefly describes a neural network as interconnected adaptive units…
This notebook demonstrates a portfolio optimization workflow using FinRL’s PortfolioOptimizationEnv and the EIIE policy architecture. It downloads data for ten Brazilian stocks, scales each stock’s series, trains a policy-gradient agent on an earlier period,…
This BigQuant help response clarifies which data fields are available in its visual automatic stock-labeling module. It says the module directly supports only the standard open, high, low, and close price fields. A user asking about unavailable documentation…
This code module implements three neural-network architectures that could be applied to quantitative prediction tasks: a feed-forward multilayer perceptron, an LSTM-based recurrent network for sequential inputs, and a Pi-Sigma network that multiplies…
The document asks how to use a user-uploaded file as a feature in an LSTM trading strategy. The suggested approach is to pass the uploaded data through a data-source module and join it with the strategy’s existing data, making the added values available…