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
A trader reports that a stock-selection strategy using TA-Lib indicators produces different results each time it is run. The response points to an SQL query in one module that lacks an explicit sort order as the likely cause. When rows arrive in an…
This article proposes a compact framework for reviewing the equity market by sector rather than inspecting every stock individually. To gauge sector persistence, it counts stocks moving from an initial limit-up session to a second consecutive limit-up…
This research summary describes a study of retail investor attention and asymmetric volatility in China’s A-share market. The authors use the balance of positive, negative, and neutral posts on online stock message boards to construct a proxy for asymmetry…
This user post describes rolling model training for stock forecasts using two approaches. The StockRanker example updates the model annually, fitting on one year of data and predicting over the following year. The XGBoost example instead constructs monthly…
This Chinese equity research excerpt analyzes the beer industry’s shift from pursuing output scale toward improving profitability through premium products. It separates the outlook into volume and price per ton: production depends on the number of operating…
This article examines how training-window length affects an AI stock-selection model. It compares long windows, expanded year by year from 2005 to 2021, with shorter windows ranging from a month to a few years. The author recommends checking each window…
This discussion addresses how to use factors from BigQuant's factor library as model features. The question asks whether a factor identifier can be used directly or whether users must copy its expression, and whether factors associated with a brokerage are…
This article introduces Granger causality and transfer entropy as ways to study directional predictive relationships in time series. Granger causality tests whether a source series improves forecasts of a target beyond the target’s own history, commonly…
This article explains a proposed arbitrage mechanism involving Chinese A-shares, stock purchases, securities borrowing, and short sales. Its example describes buying shares to push the price higher, borrowing shares to sell at that higher price, and later…
This Chinese-language note describes a stock prediction exercise using four cross-sectional inputs: total market capitalization, turnover, trailing price-to-earnings, and quarterly net-profit year-over-year growth. It compares linear regression, XGBoost, and…
This document summarizes an introductory treatment of graph neural networks (GNNs), which learn from data represented as entities and their relationships. It explains how convolutional ideas used with images can be generalized to graph structures, whose…
The article reviews the long decline in mutual fund and ETF expense ratios and links it to investor choices, competition, and changes in financial advice and retirement plans. It distinguishes asset-weighted averages, which reflect where investors put their…
This research review summarizes a study of manager changes at Chinese equity funds, distinguishing internal transfers, moves to another public fund, moves to a private fund, and exits from the fund industry. The study compares benchmark-adjusted buy-and-hold…
The article introduces a custom coded stock selection strategy built from five rules, mainly intended to identify stocks that have fallen sharply and begun to stabilize, with some rules meant to capture potential strong advances. It contrasts AI generated…
The document asks whether minute or second level price and volume data can be aggregated into daily net flow measures for very large, large, medium, and small trades. It points readers to two implementations: a general set of high frequency factors and a…
The document explains gold contracts for difference as cash-settled derivatives that provide exposure to gold price changes without holding bullion or futures. It outlines features such as leverage, smaller contract sizes, the ability to take long or short…
This research review examines momentum in Chinese equities and compares several ways to measure it. It reports that simple one-year returns, with a recent period excluded, did not predict returns reliably across the full A-share market in the tested sample.…
This research summary examines the difference between monthly stock and industry behavior in Chinese equities: individual stocks are described as showing reversal, while industry indices show momentum. It proposes an industry-momentum factor based on the…
This event description focuses on why quantitative strategy results in backtests can differ from live trading performance, a concern it identifies specifically for Chinese A-share strategies. It presents the topic as a practical challenge for traders and…
This article recommends a three-part process for improving trading discipline: record each trade’s rationale, timing, size, planned risk and reward, and emotional state; review those records regularly for rule violations and emotional decisions; then analyze…
The article argues that trading researchers should judge machine learning papers by simplicity, reproducibility, and generality, alongside reported empirical results. A method that performs well on a benchmark may not transfer to financial prediction, and a…
This study compares ARIMA, a multilayer perceptron, an LSTM, and an ARIMA-GARCH combination for next-day NVIDIA share-price forecasts. It describes ARIMA as a model for linear time-series structure, neural networks as ways to capture nonlinear patterns, and…
This interview discusses how artificial intelligence, large datasets, and cloud computing support one another: AI methods analyze data, while cloud infrastructure supplies scalable storage and computing. It maps the data industry across collection, storage,…
The report proposes improving traditional liquidity factors for Chinese A-share stock selection by incorporating information about intraday trading structure. It frames the approach as a reusable factor research method: identify market information omitted by…