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.
22,592 documents
This forum post raises a factor-data debugging question: why can an industry-neutralized price-to-book factor contain missing values when the raw factor appears complete? It identifies an industry-neutralization function applied using a first-level industry…
This Chinese-equity screening proposal combines alignment among at least five moving averages with a condition that the 20-day average exceeds the 120-day average. It also calls for considering company characteristics, including industry context,…
This note describes a simple Chinese-equity screen using three conditions: daily high-low range relative to the prior close above a threshold, closing price above its five-day moving average, and a closing price of 18.5 yuan. The accompanying explanation…
This report surveys hedging choices for U.S. equity portfolios during drawdowns and recessions, drawing on a historical sample from 1985 through 2018. It contrasts rolling index puts and short credit risk with Treasury and gold exposures, then examines…
This review compares passive and active approaches to protecting U.S. equity portfolios during severe drawdowns and recessions. Its sample covers 1985–2018, including eight equity declines greater than 15% and three U.S. recessions. It discusses rolling S&P…
The post proposes a Chinese equity screen combining daily price amplitude above 1%, a robotics concept classification, circulating market value below 10 billion, and selected beverage or alcohol related industries. It outlines the conditions in prose and…
This stock screen looks for a daily range above 1%, rising bottoms, and an opening price within 5% of the 10-day moving average. The intended logic is to combine active price movement with improving support and an opening level near a short-term trend…
This stock-screening proposal combines a daily high-low range greater than 1%, a large-order net-flow measure, and a KDJ K-value change. It describes ranking candidates by KDJ change and frames the approach as a short- or medium-term technical screen. The…
This stock screen combines three filters: membership in the metaverse theme, a previous-day price at or above the 250-day moving average, and an external-to-internal trade volume ratio above 1.3. The article interprets the long moving average as a trend…
The document describes a Chinese equity screening idea that ranks stocks by capital-flow strength, then selects companies with no losses, market capitalization below a stated threshold, and more than two limit-up sessions in the recent ten-day window. It…
This article proposes a qualitative decision framework based on comparing broad market strength with the strength of individual stocks. It organizes conditions into four cases: when both market and stock are strong, favor stock exposure; when the market is…
This document presents a simple Chinese equity screening rule: select stocks with turnover between 3% and 12%, circulating market capitalization between 5 billion and 10 billion yuan, and a positive return. It gives both a formula-style expression and a…
This notebook outlines a machine learning workflow for daily CSI 300 constituent stocks. It loads historical bars and changing index membership filters, constructs an Alpha101 dataset, and divides the sample into training, validation, and test periods. The…
This Chinese equity screening idea combines price amplitude above 1%, appearance on the prior day’s trading list, and return on equity above 15% for five consecutive years. The post frames these conditions as a blend of short-term volatility, market…
This article argues that LSTM models trained on historical prices can appear to forecast stocks accurately because predicted daily prices often remain close to the previous day’s price. When plotted against actual prices, this persistence can look…
This post outlines a Chinese equity screen combining daily price amplitude above 1%, a 2021 date condition, and three consecutive limit-up sessions. Its final stated criteria also require trailing price-to-earnings below 30 and price-to-book below 5. The…
This documentation explains how to use Polygon as a historical price-data source for LumiBot backtests across stocks, options, forex, and cryptocurrencies. It describes supplying an API key, selecting a backtest date range, and running a simple example…
This Chinese-language post proposes screening stocks with at least five moving averages converging, an outstanding convertible-bond name, and a price gain below 6% at 9:25. It describes the moving-average condition as a way to find stocks with a potentially…
This Chinese stock-selection example targets companies in the metaverse industry whose opening price is near or above a ten-day moving average and whose reported daily position growth exceeds five percent. The article frames the moving-average condition as a…
This Chinese stock-screening example combines three short-term filters: a daily high-low range above one percent, a large-order net-flow measure above a stated threshold, and an auction-related price change between negative two and positive five percent. It…
This proposed stock-selection method combines three conditions: a historical dividend payout ratio above a stated cutoff, an opening price near the ten-day moving average, and at least five moving averages described as overlapping. The rationale is to find…
This Chinese-market stock screen selects shares whose daily high-low range exceeds a threshold, that did not close at the upper price limit on the previous day, and whose current-day return is positive. The stated rationale is to avoid very quiet stocks,…
Williams %R measures a close against the recent period’s high and low. Its scale runs from −100, when the close matches the period low, to 0, when it matches the period high. The document notes that the oscillator can change quickly, so signals based on its…
This post describes a Chinese A-share screening rule that combines a daily high-low range above one percent, a date filter for 2021, and a condition intended to identify a prior session's 9:15 matched price at the limit-down level. It sketches equivalent…