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 example describes an AI trading workflow for leveraged long and inverse ETFs. A researcher ranks the available funds, separate bull and bear agents present competing cases, and an interpreter turns their views into target account weights. A trading…
The document summarizes research applying Wasserstein generative adversarial networks (WGANs) to financial return series and comparing them with standard GANs. WGANs replace the original GAN’s distribution-distance measure with Wasserstein distance, which…
This configuration describes a Qlib workflow for training a LightGBM model on Alpha360 features and evaluating its predictions on CSI 300 constituents. The label compares closing prices across the next two reference periods. The data handler drops missing…
This configuration specifies a Chinese equities prediction and portfolio backtest using Qlib, LightGBM, and the Alpha158 feature handler. It pairs daily labels with one-minute features, resampling the minute data at 14:56. The listed data span begins in 2008…
This article recommends five books as an entry path into quantitative and algorithmic trading. It starts with accessible overviews of retail systematic trading and professional quant funds, then points readers toward execution algorithms and exchange…
This Chinese-language summary outlines an equity market-timing approach based on a long short-term memory (LSTM) neural network. The model is described as learning dependencies across time-series observations and adapting to sharp changes in trends. It…
This introductory guide describes a workflow for building an automated cryptocurrency strategy with OctoBot Script. It explains reinforcement learning in broad terms: an agent interacts with a market environment, takes actions such as buying or selling, and…
The document outlines a beginner workflow for creating an AI-assisted stock selection strategy on a quantitative platform. It starts with defining a trading style, translating that idea into candidate features, expressing those features for model training,…
This overview groups Python packages by common data science tasks. It covers core numerical and statistical tools for arrays, data manipulation, model estimation, and hypothesis testing; plotting libraries for static and interactive charts; and machine…
This Qlib configuration defines a temporal convolutional network workflow for ranking CSI 300 stocks. It uses Alpha360 features, robust feature normalization, missing-value filling, and cross-sectional rank normalization of labels. The prediction target is…
This research summary tests whether negative sentiment in posts on Xueqiu, a Chinese investing forum, can help time major Chinese equity indexes. It describes using machine learning to classify post sentiment and building a weekly negative-sentiment ratio…
This brief support exchange asks how the numbers displayed after decision branches in a visual stockRanker model are calculated. The response points the reader toward material on tree models and information entropy, implying that the displayed values relate…
The article introduces quantitative trading as the use of mathematical models, statistical analysis, historical and real-time market data, and algorithms to guide or automate decisions. It describes stock screening and rule-based execution as common…
This brief support exchange addresses a user’s report that a hyperparameter module raises a key error even though the key appears in the visual chart and can be found in code-mode parameters. The response attributes the issue to an error in the…
This brief describes a deep neural network approach intended for quantitative stock selection. Its stated implementation goals are to run on systems with limited memory, use GPU acceleration when available, and adapt the number of input features. It also…
This meetup Q&A surveys several quantitative trading topics: using Level 2 order-book data, reproducing a legacy strategy, allocating idle capital to a gold ETF, factor decomposition, building CNN strategies, and portfolio optimization. Most answers point…
The document introduces the task of developing a complete strategy from daily factors extracted with a high-frequency expression engine. It identifies strategy development as the topic of a meetup presentation and points readers to a related recording, but…
This tutorial builds a supervised learning strategy for hourly EUR/USD data. It derives features from moving averages, momentum, Bollinger bands, a synthetic sentiment signal, and time of day. The target class represents whether the price return roughly two…
This guide surveys methods for improving and checking machine learning models: sensitivity analysis, residual diagnostics, baseline comparisons, security review, data augmentation, model editing and monitoring, output assertions, and anomaly detection. It…
This strategy combines a restricted China A-share universe, model-based ranking, technical timing, and portfolio controls. It starts from a manually selected group of roughly 100 to 300 large-cap constituents associated with the CSI 150 universe. A…
This article compares neural networks, decision trees, random forests, support vector machines, and a stock-ranking algorithm for equity selection. It describes broad trade-offs: neural networks can model nonlinear relationships and varied data structures,…
This educational-session overview introduces machine-learning and deep-learning approaches to quantitative stock selection. It lists core workflow topics: defining prediction labels, engineering features, separating training, validation, and test sets, and…
The document introduces a periodically rolling retraining framework for forecasting models. Its central idea is to refresh a model using up-to-date data at intervals so its forecasts can adapt as market conditions change. The default model is linear, and…
This study compares loss functions for neural models that forecast five-day cumulative returns for selected Chinese A-shares. It uses a DeepAlpha DNN as the main model and compares mean squared error with absolute error, Pseudo-Huber, negative information…