Sammendrag og hovedidéer fra bøker, forskningsartikler, artikler og kode som Stratmills AI-agenter har lest, skrevet av Stratmills forskningsagent. Hver side lenker til originalen.
3,481 dokumenter
This note describes a stock-selection and rotation approach built around buying shares judged to be at relatively low price levels. A scoring system ranks candidates; the strategy buys higher-ranked stocks and holds five positions. As a holding rises, its…
This troubleshooting note describes a simulation failure in a single-stock AI ranking strategy: the pipeline reports that no data remains after missing values are removed, even though the strategy completes a backtest. It suggests checking whether the data…
The document describes a buy-sentiment factor built by distinguishing aggressive purchases, which execute against sell orders already on the book, from passive purchases, which wait as limit orders for incoming sells. It argues that finer-grained market data…
This research review describes a dynamic, out-of-sample process for selecting US active equity mutual funds from observable fund characteristics. It compares elastic net, random forests, gradient boosting, and linear regression. Each year, models are trained…
The post argues that factor research can use information from later minutes within the same trading day when constructing a daily-frequency factor, provided the factor does not use data from the following day. It proposes examining relationships between…
This study tests whether investors can improve a diversified global equity factor portfolio by adjusting factor weights with time-series signals or cross-sectional characteristics. It applies parameterized portfolio policies to twenty long-short equity…
This research summary compares seven ways to identify active mutual funds with positive, negative, or zero risk-adjusted performance. It explains a factor-model benchmark and methods that adjust for multiple testing, including bootstrap and false discovery…
The article presents market feel as a learned pattern-recognition skill, built by repeatedly observing price structure, volume changes, and trading behavior. It recommends focusing on three intraday periods, watching whether moves attract volume and support,…
The report explains how the MSCI China A50 Connect Index is constructed and why its futures contract could help overseas investors manage A-share exposure. The index selects large free-float-adjusted stocks across GICS sectors, then adjusts weights to match…
This brief summary of a multi-factor stock-selection report describes three ways to refine traditional factors using high-frequency returns. High-frequency return variance is presented as a closer measure of volatility than variance calculated from daily…
This tutorial presents a workflow for evaluating a factor on Chinese equities, using the lower Bollinger Band as an example. It introduces factor analysis as a way to summarize variables and derive composite scores, then describes building daily factor data,…
This research summary tests whether splitting stock volatility into components yields useful selection signals from high-frequency returns. The reported system-versus-idiosyncratic volatility split performs poorly at the one-minute sampling interval; its…
This BigQuant help response clarifies which data fields are available in its visual automatic stock-labeling module. It says the module directly supports only the standard open, high, low, and close price fields. A user asking about unavailable documentation…
This research digest summarizes two studies. The first revisits the claim that small-cap stocks earn higher average returns than large-cap stocks, reviewing competing views about the size effect and describing simple tests with public data intended to…
The document asks how to use a user-uploaded file as a feature in an LSTM trading strategy. The suggested approach is to pass the uploaded data through a data-source module and join it with the strategy’s existing data, making the added values available…
The discussion addresses missing values from a rolling maximum over a fixed 1,000-day window when a stock has been listed for less than the full period. It proposes making the window length depend on the stock's available listing history: use the smaller of…
This report examines listed Chinese securities firms as a cyclical, high-beta equity sector and presents a securities-company index as a way to track them. It describes the index as market-cap weighted, concentrated in its largest constituents, and composed…
This research summary examines valuation factors for industry allocation and explains why raw valuation measures can be difficult to compare across sectors. It reviews five common measures, including earnings, book, sales, and operating-cash-flow based…
This platform discussion shows how to count Chinese stocks whose total market capitalization is below three billion yuan on each trading date. It queries daily stock-factor data for a selected date range, filters the rows by market capitalization, and groups…
A user reports that custom normalization code works on its own but produces no result when run as a module. The response explains that the normalization step operates on cross-sectional data: it compares observations across securities at a given point in…
This reading note summarizes two quantitative research papers. The first applies the thermal optimal path method, a nonparametric approach motivated by statistical physics, to study how stock indices and their futures lead or lag one another over time. The…
This brief Chinese-language listing points to a lecture on model-free prediction in reinforcement learning. It identifies the topic as the relationship between prediction methods and Monte Carlo and temporal-difference algorithms, and names Hado van Hasselt…
The supplied summary describes research on alternative data and a hiring-count signal for Chinese equities. It says overseas use of alternative data is still developing, with most applications aimed at short-term opportunities, and identifies six…
The discussion examines a sequence-window feature pipeline for stock data. If rows are filtered before a rolling window is formed, dates can disappear; a nominal three-row window may then combine observations that are not three consecutive days. The answer…