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.
4,194 documents
This tutorial introduces gplearn, a Python library that uses genetic programming for regression and symbolic regression. Candidate expressions are iteratively selected, crossed, and mutated to find mathematical relationships in data. The resulting formulas…
The document describes a screening rule for Shenzhen main-board stocks that combines a daily amplitude threshold above 1% with a closing price below 20, positive price-to-earnings below 29.01, and positive price-to-book below 3.11. It includes formula and…
This study examines value, momentum, carry, and defensive factor returns across six asset classes using data spanning roughly a century. It compares each factor’s discovery period with earlier and later periods, using the long history and cross-asset…
This stock-selection recipe combines a turnover-rate band of 3% to 12%, membership in a beverage and alcohol category, and return on equity above 15% for five consecutive years. The stated rationale is to pair current trading activity with industry…
The document reviews a holdings-based framework for attributing Spanish domestic equity mutual fund performance to four sources: momentum-related trading, security selection, active market timing, and passive timing effects caused by changing security betas.…
This A-share strategy ranks eligible stocks using year-over-year net-profit growth and total market capitalization. It subtracts each stock’s size rank from its growth rank, favoring companies with stronger earnings growth and relatively smaller market…
The proposed screen combines four conditions for stocks: daily high-to-low amplitude above one percent, membership in a robotics concept group, circulating market capitalization below 10 billion yuan, and a positive price-to-earnings ratio. The article…
This Chinese A-share screening idea combines a price-range condition, a prior-day price-limit exclusion, and membership in a metaverse concept group. Its stated purpose is to find stocks with notable trading activity that are associated with an emerging…
This report summary reviews Schroders as an asset manager and describes its investment approach, asset mix, and historical fund performance. It says the firm focuses on asset management, with equities prominent in its portfolios, and reports that many funds…
This document presents a rules-based screen for Shenzhen main-board equities. It combines daily price amplitude with a historical dividend-yield condition and bounded price-to-earnings and price-to-book ratios. The intended result is a list of companies…
The document presents a stock screening idea that combines daily price amplitude, a positive weekly MACD signal, and company characteristics. It first describes company type as an ambiguous criterion, then proposes making the fundamental screen more specific…
The document describes a Chinese stock screen using turnover between 3% and 12%, year-over-year net profit growth attributable to parent-company shareholders between 20% and 100%, and a stock-code prefix condition. It provides example implementations for a…
This note presents an investment case for China’s technology sector and describes the CSI Technology 50 Index. It argues that reduced access to overseas technology through acquisitions, combined with domestic innovation policies, creates a basis for…
This report summary presents a strategic allocation framework that shifts the budgeting unit from asset classes to macroeconomic drivers, aiming to reduce exposure to changing correlations among assets. Principal component analysis is used to derive five…
The document reports a factor built from the standardized ratio of closing price to its 20-period average, divided by price-to-book, then normalized by its 60-period standard deviation. It gives an information coefficient of -0.02 and reports cumulative…
This research replication constructs a Chinese equity factor called “Flowers in the Forest” by calculating three component factors and combining them with equal weights. The source describes deriving the daily factor from high-frequency data, a…
The document lists fitness measures for a genetic programming module that evolves stock factors. It describes ICIR as the mean information coefficient divided by its standard deviation, with the information coefficient measuring the relationship between…
The post describes an equity screen that combines daily price range, positive but capped ten-day appreciation, and evidence of increased institutional holdings. Its rationale is that range can indicate movement, while a moderate recent gain may identify…
This report tracks three China-focused active equity quant portfolios against an active mutual fund benchmark: an enhancement portfolio built from selected fund holdings, a portfolio that filters earnings-surprise stocks, and an enhancement portfolio based…
The document describes an equity screen for Shanghai-listed stocks. It selects companies whose turnover rate falls within a stated band and whose return on equity exceeded a stated threshold over each of the recent five years. The rationale is to combine a…
This study summary describes a low-risk portfolio approach for South Korean equities. It forecasts KOSPI 200 constituent volatility with Gaussian process regression, support vector regression, an artificial neural network, and GARCH, then ranks stocks into…
This stock-selection approach combines a turnover filter, growth in net profit attributable to shareholders of the parent company, and a technical condition described as moving averages spreading upward. The stated thresholds are turnover between 3% and 12%…
This stock-selection approach combines a price-range condition with profitability and trend filters. It selects stocks whose recent high-low range exceeds a volatility measure, whose return on equity has stayed above the stated threshold for five years, and…
This competition update reports diagnostics for the highest-ranked factors and tracks a CSI 1000 index enhancement portfolio. Factor descriptions are anonymized: the report gives average style exposures estimated through daily multivariate regressions and…