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 support note explains a model-training failure attributed to problems in the input data. The specific diagnosis is that an erroneous expression caused the buy-condition factor to fail extraction. It also notes that the dataset had only two factors, both…
This BigQuant example builds a binary classifier from rolling windows of daily open, high, low, close, and volume changes. Its network applies a convolutional layer to each 50-step, five-feature input, passes the result through an LSTM and dense layers with…
The document introduces attention and Transformer components, then adapts them to forecast stock returns from factor sequences. Its proposed architecture adds Time2Vec features to factor inputs, applies self-attention and encoder layers, and uses a…
This notebook demonstrates an end-to-end Qlib workflow for a China-market stock ranking model. It initializes Qlib data, uses the CSI 300 universe and Alpha158 features, and trains a LightGBM model on historical data with separate training, validation, and…
This document introduces an MQL5 port of ALGLIB, a cross-platform numerical analysis library, and outlines how its packages can support quantitative research. Covered methods include linear algebra, equation solving, interpolation, optimization, Fourier…
The document introduces decision trees for classification and regression. It explains how a tree recursively partitions observations using feature tests, with branches representing outcomes and leaves providing class predictions or numeric estimates. Tree…
The document explains how to bring financial statement and macroeconomic data into a quantitative model as features. It distinguishes macro data, which is readable from the platform but must first be aligned and merged with the sample data, from financial…
This brief platform support exchange distinguishes missing feature values from NaN values produced by calculations. The support response says missing values had already been filtered; remaining NaNs were calculation results, such as an undefined zero divided…
The document summarizes a study of Alpha101 factors in China’s A-share market and a method for generating additional price-volume signals with genetic programming. It says the authors assessed existing factors using Rank IC, portfolio stratification, and…
This FAQ describes the scope and practical use of an educational financial reinforcement learning library. It covers supported data sources, feature inputs such as sentiment, training options, reward functions, hyperparameter tuning, and algorithm choices.…
This documentation entry directs coding agents to start from complete LumiBot examples for either AI-based or ordinary Python strategies. It describes the strategy lifecycle at a high level: create agents during initialization and invoke them during each…
This forum post raises a data-preparation problem in an LSTM trading strategy. The author reports that filtering rows before rolling sequence windows can remove observations that should remain part of the temporal context, causing later observations to shift…
The article compares two ways to build equity strategies: train a predictive model on chosen factors and a defined stock universe, or specify screening conditions directly. The AI approach can be quicker to assemble, but its behavior may be hard to diagnose…
The paper examines how five emotions expressed on Sina Weibo—anger, disgust, fear, joy, and sadness—relate to Chinese stock returns. It uses machine learning to classify posts and wavelet analysis to study how the relationship changes over time and across…
This competition entry modifies a Deep Alpha convolutional neural network by supplying five time slices for each of 98 factors, rather than using a single slice. The network structure and most parameters remain close to the original model because training…
This overview summarizes six ICML 2022 papers selected by G-Research staff. The topics include convex formulations for training two-layer ReLU networks, linear-time attention, branch-and-bound optimization of shallow decision trees, structured matrices for…
This comparison surveys AI-oriented trading and research projects by their agent workflows, ability to replay or backtest decisions, broker paths, deterministic strategy support, and hosting or monitoring features. It distinguishes research-focused tools…
This post sketches an experimental model for identifying leading stocks, using gradient boosted regression to score shares within a rolling window. The proposed label gives higher scores to stocks that rise more, experience smaller drawdowns, and break out…
This comparison explains how Lumibot and QuantConnect LEAN differ as algorithmic trading frameworks. Lumibot is presented as a Python-first library in which strategies use ordinary Python classes, broker and data adapters, and can combine deterministic rules…
DoubleEnsemble is an ensemble framework for financial prediction that addresses noisy training data and large feature sets. It combines two procedures: it uses each sample’s learning trajectory during training to identify influential examples and adjust…
The document describes a hybrid approach that combines ARMA-GARCH models with several neural network types to identify market patterns and predict the direction of shocks in defense equities and foreign exchange. ARMA-GARCH models capture linear time-series…
This overview introduces reinforcement learning as a way to learn sequential decisions by interacting with an environment and maximizing accumulated reward. It outlines the agent, environment, policy, and reward, and explains how delayed feedback differs…
The document introduces reinforcement learning through the interaction of an agent and an environment, where actions produce feedback and the objective is to maximize discounted future rewards. It frames the problem as a Markov decision process, explaining…
This overview describes several ways quantitative trading attempts to identify and capture market opportunities. It presents historical data analysis and backtesting as tools for assessing a strategy, then discusses automated execution, momentum examples,…