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.
3,481 documents
This stock-selection rule screens for shares with daily amplitude above 1%, a tradable share count no greater than 5.5 billion, and at least one daily gain of 10% or more during the prior 25 trading days. The article presents these filters as a way to find…
This note explains why Chinese stock prices from a platform's basic feature extraction may differ from values returned by its data-source API. The example concerns Sany Heavy Industry and compares API output with feature extraction output. The explanation is…
This BigQuant question addresses how to identify a Chinese stock’s daily upper and lower price limits when floating-point calculations differ from exchange prices. The suggested method is to read the platform’s limit-price data source for Chinese A-share…
This research note extends a spectral approximation for covariance matrices used in portfolio optimization. It derives a concise upper bound for approximation error and uses that bound to choose the number of retained eigenvalues dynamically. The stated aim…
This practitioner note argues that both feature selection and model choice shape quantitative strategy performance. It compares Random Forest, scikit-learn gradient boosting and random forests, and XGBoost, describing tradeoffs in robustness, speed,…
This study compares equity portfolios selected on profitability, growth, and cash flow, then tests combinations of these fundamentals. It reports quarterly rebalanced results against the CSI 300 from 2013 through June 2021. Each standalone top-50 factor…
This article proposes reading a stock’s first half-hour of trading relative to its opening price. It gives five patterns: a pullback that holds above the open, a rally that falls below it, a rebound that fails to reclaim it, a rebound that breaks and holds…
This brief project note describes using small market capitalization as a factor and training a model to relate that factor to subsequent 10-day returns. It presents an initial quantitative research idea: use a size characteristic as an input and forward…
This short BigQuant forum post concerns a backtest error indicating that no data remains after missing values are dropped. The suggested diagnosis is that stock-filter settings may be too restrictive, removing every candidate before the backtest proceeds. In…
The report outlines a fundamental framework for selecting Chinese technology and innovation stocks. It first identifies candidate industries through subjective classification or quantitative screening based on lower sensitivity to macroeconomic fluctuations,…
This research tests whether lagged returns in economically related industries can help predict future industry returns, based on the idea that information and shocks may diffuse across sectors over time. To limit overfitting in a large predictor set, its…
This Chinese-language research digest summarizes two topics. The first concerns systematic style investing in fixed income: it reports that value, momentum, carry, and defensive premia have shown notable results in both government and corporate bonds,…
The article describes selecting Chinese “monster stock” reversal candidates with a changing, multi-factor score. It examines recent profitable picks by industry and a normalized factor, then uses those observations to infer which sectors and factor ranges…
This brief guide describes a template for trading exchange-traded funds with MACD crossovers. Construct separate factors for bullish and bearish MACD crosses, then use the bullish cross as a buy signal and the bearish cross as a sell signal. It notes that…
The note translates a discretionary technical approach into simple stock-selection rules. It excludes certain Chinese listing boards, ST-designated stocks, and stocks below or above stated market-cap limits. Entry requires price above the five-day moving…
This research summary describes applying XGBoost, a gradient-boosted tree method, to Chinese stock selection and index enhancement. The model uses CART trees and classifies each stock into one of three groups formed by sorting market returns. This turns…
This research summary describes a method for extracting nonlinear relationships between equity style factors and returns. It keeps an interpretable linear factor model as a baseline, then trains machine learning models on the linear model’s residuals. The…
This forum post raises a feature-engineering question: why a named 14-period RSI feature, a function-based RSI expression, and a manually written RSI formula return different values. The user describes extracting all three through default basic-feature and…
This article introduces graph neural networks and focuses on graph attention networks (GATs) for financial time-series prediction. It explains that GATs use attention to weight neighboring nodes' features, allowing the weights to depend on node data rather…
This document summarizes two discussions of portfolio diversification. The first describes Bridgewater’s explanation of risk parity and its All Weather strategy, including a comparison with traditional portfolios and responses to common criticisms. The…
This Chinese-language research digest discusses a common source of backtest overfitting: researchers run many experiments, then report only the strategy or factor with the strongest historical result. Selecting a winner from repeated trials can make apparent…
This Chinese-language forum exchange discusses transaction costs and bar frequency in an event-driven backtesting platform. A respondent says that sell-side stamp duty can be represented as an added transaction fee, while the platform’s built-in frequencies…
This factor study examines stock selection in Hong Kong equities and contrasts the market with mainland China. It describes Hong Kong as having a more polarized company-size distribution and lower trading activity, then tests market-cap, momentum, growth,…
This 2020 weekly note reviews Chinese equity market behavior and considers which investment styles may be relevant during earnings-report season. It combines calendar patterns, market dispersion, fund flows, valuation levels, and macroeconomic events to…