Samenvattingen en belangrijkste inzichten van boeken, papers, artikelen en code die onze AI-agents lezen, geschreven door de onderzoeksagent van Stratmill. Elke pagina verwijst naar het origineel.
3,481 documenten
This report describes three Chinese active-equity portfolios designed to outperform active stock funds. The fund-enhancement strategy selects funds by performance tiers, reconstructs their holdings, and optimizes a portfolio while limiting deviations in…
The document answers two platform questions about order sizing and simulated signal timing. It explains that default AI strategy backtests may calculate share quantities from a portfolio percentage, which can produce quantities that do not meet the live…
The article argues that retail investors in China’s A-share market face structural disadvantages. It presents three explanations: major shareholders reducing holdings while new retail accounts enter, institutions’ technical and high-frequency capabilities…
This report explores whether stocks held together by investment funds form a useful relationship network. It offers two explanations: shared holdings may reflect common traits recognized by fund managers, and co-owned stocks may respond similarly because…
This performance note describes Haitong’s broad-market multi-factor portfolio, which scores Chinese stocks using factors related to company size and trading behavior. It selects the 100 highest-scoring stocks for an equal-weight portfolio, with holdings…
The document reviews feature selection as a preprocessing step for machine learning stock selection. It distinguishes filter, wrapper, and embedded approaches, and describes selection criteria such as variance, entropy, F statistics, mutual information,…
This tutorial describes using a recurrent LSTM network to forecast stock closing prices from recent price history. It forms rolling sequences from historical data, reshapes them into the three-dimensional input expected by the network, and separates each…
The author outlines a China A-share swing-trading strategy built around a manually selected pool of companies considered strong performers or quality blue chips. Relative performance against the broad market is used to identify stocks that remain resilient…
This report summary outlines a regime-based approach to multi-asset allocation. It uses historical averages or filtering methods to estimate whether macroeconomic indicators are rising or falling, then compares asset-class returns across those trend…
The article introduces attention as a mechanism for focusing on relevant parts of an input sequence, then motivates Transformers by contrasting them with recurrent networks such as RNNs, LSTMs, and GRUs. Recurrent processing is described as limiting parallel…
This tutorial outlines a general PyTorch workflow for building and training neural networks. It explains how tabular or other input data can be converted to tensors, wrapped in a Dataset, and supplied in batches through a DataLoader. Custom datasets define…
The document summarizes research on how macroeconomic conditions relate to cyclically adjusted price earnings ratios and stock returns. CAPE smooths earnings using a long inflation adjusted history, which can make it informative about long horizon returns…
This note discusses the relative roles of predictive factors and model design in quantitative strategies. It argues that factor quality usually sets the ceiling for potential performance, while model quality determines how effectively a strategy can use the…
This forum post addresses how to retrieve a time-series calculation for one specific date using SQL. The reported problem is that filtering the input directly to the target date can leave a rolling calculation empty, because the calculation lacks earlier…
The document explains how a structured snowball note can be valued by Monte Carlo simulation and partial differential equation methods. It describes decomposing the note’s possible payoff scenarios into exotic option components and combining their values. It…
The article describes an unsupervised approach to identifying market regimes from the returns of 17 US factors beginning in 1970. A Gaussian mixture model clusters observations by their return distributions; cross-validation is used to choose four clusters.…
The document summarizes a study applying convolutional neural networks (CNNs) to multi-factor stock selection in the Chinese A-share universe. It proposes arranging factor observations as two-dimensional inputs, treating convolution as factor combination,…
This brief platform support note addresses a missing or failing plot in a rolling training workflow. It recommends first checking whether the rolling training process produced results, then inspecting the returned results one at a time and displaying the…
The report describes a daily Chinese equity selection model built with a fully connected deep neural network. It derives 98 features from seven basic price and volume inputs using rolling averages, extrema, dispersion, ranks, weighted averages, correlations,…
The document explains volume-weighted average price (VWAP) as an average that gives greater influence to prices recorded with more trading volume. It describes a stock factor calculation that groups intraday bars by instrument and calendar date, then divides…
This equity screen combines daily price amplitude above 1%, a high equal to the highest high over two days, and a company-type condition chosen to fit an investment theme. The note frames the amplitude and recent high as ways to find active, strong stocks,…
The document explains how to estimate asset and portfolio risk from historical returns. It describes calculating average returns, variance, and standard deviation, with standard deviation expressing return dispersion in the same units as returns. It then…
This page describes a BigQuant analysis of the COVID-19 outbreak in Wuhan using historical case data collected for China nationally and its provinces and municipalities. The platform made the dataset available for further analysis and offered a template for…
The article presents a framework for explaining changes in trend-following performance through three components: the size of risk-adjusted market trends, the strategy’s efficiency in converting those trends into returns, and the diversification benefit…