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 post asks how to define groups for a learning-to-rank model in a stock selection setting. In ranking algorithms, each group contains observations that should be ranked against one another; the author proposes grouping stock samples by trading date so…
This indicator gathers values from up to ten combinations of symbols, timeframes, and signal variants, synchronizing them to the symbol on which the indicator is installed. Its original purpose is to provide normalized inputs to a neural network, though the…
FactorVAE is described as a probabilistic dynamic factor model that applies a variational autoencoder to noisy market data to forecast cross-sectional stock returns. In training, an encoder-decoder uses future returns to identify posterior factors that…
FactorVAE is presented as a probabilistic dynamic factor model that uses a variational autoencoder to extract latent factors from noisy market observations and forecast cross-sectional stock returns. During training, an encoder and decoder use realized…
The document motivates adapting forecasting models to changing market conditions: financial data distributions can shift over time, so models trained on earlier periods may lose predictive strength. It compares two approaches, RR and DDG-DA, using linear and…
This note compares portfolio construction approaches, including equal weighting, market-cap weighting, and stratified sampling, with optimization methods such as mean-variance optimization, risk parity, target-risk, ABL, and tracking-error models. It…
This brief Chinese-language forum exchange concerns whether a user can retrieve feature importance from a random forest in a visual quantitative-strategy platform. The reply identifies the random forest module and suggests creating a visual strategy template…
This research summary studies whether lagged returns from related industries can help predict industry returns for sector rotation. It motivates the question by noting that such predictability would not be expected in an ideal frictionless market, while real…
The article outlines common aims of algorithmic execution: splitting a parent order to follow market volume, reducing market impact, and concealing trading intent. It also presents automation as a way to improve order-entry efficiency and reduce manual…
This documentation explains how to configure FreqAI within a Freqtrade configuration and strategy. It outlines core settings for training and backtesting periods, model identification, timeframes, correlated pairs, shifted candles, indicator periods, labels,…
This forum post describes a reproducibility question in a quantitative modeling workflow. The author sets a learning-rate parameter to the same value in the model and in a hyperparameter-search module, and configures the search grid to contain only that…
This introductory explanation presents support vector machines as supervised classification models. In a simple two-class example, many lines can separate the data, but SVM selects a boundary that maximizes the margin between the two classes. The closest…
This document records a failure in a stock-ranking training module. The run loads 6,711 training rows, begins data and feature preparation, then stops with an IndexError while calculating maximum discounted cumulative gain. The traceback points to code that…
This document describes an adaptive trend indicator that evaluates multiple SuperTrend variants using different Average True Range multipliers. It scores their recent performance, groups the variants into three performance clusters using a k-means-inspired…
This guide introduces statistical learning through an example of predicting the S&P 500 from company fundamentals. It frames the task as estimating a response from predictor features plus residual error, then distinguishes prediction, which prioritizes…
This report overview surveys supervised learning methods with potential uses in investment research. For regression, it distinguishes penalized models such as Lasso, ridge, and elastic net, which can address correlated inputs, from nonparametric approaches…
This indicator modifies SuperTrend by grouping recent ATR observations into low, medium, and high volatility regimes with a three-cluster K-Means procedure. It initializes cluster centers from percentiles of the ATR range over a training window, repeatedly…
This Chinese-language summary introduces TS-Boost, a machine-learning framework for equity factor selection. It identifies two challenges in financial data: observations may not share the same distribution over time, and market signals can have low…
This introductory section points new users to three FinRL tutorial notebooks. One is presented as a recommended first exercise, walking through a full deep reinforcement learning workflow for stock trading. Another demonstrates connecting FinRL to Tushare…
The author describes using an AI assistant to rebuild a customized trading application from VeighNa 3.9 to 4.3. The work covered a changed modular architecture, a new database backend, a redesigned strategy data model, and a rewritten desktop interface. The…
This documentation landing page introduces ArbitrageLab, a Python library covering end-to-end pairs-trading strategies and tools for developing strategies. It organizes its subject matter around multiple approaches, including distance methods, cointegration,…
This report studies whether boosted tree models can extract stock return signals from analyst expectation factors that linear models may miss. The authors note that these factors can be complex, have substantial missing data, and correlate with earnings and…
This research note examines shifts between large-cap and small-cap leadership in Chinese equities, using the CSI 300 and ChiNext indices as proxies. It argues that rotation can reflect macroeconomic conditions, industry trends, and market sentiment, and that…
This review summarizes research spanning equity market microstructure, institutional holdings, crypto momentum, dynamic-graph forecasting, and reinforcement learning for limit-order-book trading. In equities, one study groups trades by their timing relative…