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
The note proposes adding a 20-period average true range (ATR) volatility factor to a trading strategy, with the aim of improving its Sharpe ratio. It describes a regime-dependent relationship: higher ATR is associated with stronger returns in bull markets,…
This A-share stock selection strategy buys the ten smallest companies by market capitalization each day and rebalances daily. Its universe is drawn from the CSI 1000 and excludes ST and delisted stocks. Additional filters require a positive trailing…
This BigQuant community post reports that applying a lagged value in a query that joins stock prefactors directly to an index-component table can produce zero-filled results. The example calculates a two-period stock return using a lagged close price, then…
This short forum answer addresses how to aggregate price-to-earnings data across stocks in an industry. It suggests using a grouped sum operation keyed by an industry classification and applying it to trailing-twelve-month P/E values in a stock-factor…
This article proposes interpreting value, momentum, quality, and low-volatility returns through a recurring factor cycle inferred from factor performance itself. It describes normal periods, episodes when value suffers a major drawdown due to growth stocks…
This Chinese-language summary reviews a quantitative research team’s 2017 results across four areas: multi-asset allocation, index enhancement, factor timing, and commodity trading adviser strategies. It reports that an equity-and-bond active risk-parity…
The document describes preparing supervised-learning labels from Chinese daily stock bars. It calculates a forward price ratio using lead values, clips observations to the cross-sectional 1st and 99th percentile boundaries, and assigns the result as the…
This report frames next-day direction in China’s Wind All A-share index as a binary classification task. It trains a decision-tree-based AdaBoost model on 51 daily features drawn from several markets and asset classes, including repo rates, credit spreads,…
This event page outlines a presentation on moving from discovering individual factors to designing their portfolio weights. It identifies linear and nonlinear combinations and ICIR-based weighting as methods, with attention to their underlying rationale and…
This shared open-source example builds a daily classifier for Chinese stocks using price and volume data, technical indicators, and three supervised models: a random forest, logistic regression, and a linear support vector classifier. Features include…
This summary compares two research reports on protecting portfolios during market stress: one published before a major volatility episode and another written afterward. It frames their comparison around how investors assess defensive strategies before a…
The report describes applying OpenFE, an enumerate-and-filter framework, to discover interpretable fundamental factors from Chinese companies’ balance sheets, income statements, and cash-flow statements. It first removes fields with substantial missing data,…
This research summary describes a monthly sector-rotation framework for Chinese equities. It groups 29 CITIC first-level industries into financial, cyclical, consumer, and growth sectors, builds market-cap-weighted sector indices, and proposes selecting…
This research summary describes a multi-factor approach to selecting Chinese convertible bonds. It treats each bond’s value as a combination of straight-bond value and embedded option value, and examines how bond returns relate to the underlying stock. The…
This study proposes a market mechanism for trading data products by separating data ownership from the right to use the data. Ownership is divided into tradable units to support liquidity, while use rights remain indivisible and go to an end user. The…
This overview defines factors as common drivers of security returns and distinguishes systematic factor exposure from asset-specific risk. It traces the move from the market-only CAPM to multi-factor explanations such as the Fama–French market, size, and…
The article argues that Rust can suit quantitative trading infrastructure where large data workloads, dense computation, concurrency, and low latency matter. It attributes this fit to Rust’s performance, memory and thread safety guarantees, and lack of…
The document defines the Mass indicator using the high-low range. It first applies an exponential moving average to that range, then smooths the result with a second exponential moving average of the same length. The ratio of the first smoothed range to the…
This forum post compares backtest results for a 20-day and 110-day moving-average strategy on Vanke A shares across two platforms. The strategy buys when the shorter average exceeds the longer one and exits when it falls below, using full portfolio…
This research summary describes an A-share market-impact model intended to narrow the gap between simulated strategy returns and realized portfolio returns. It argues that applying one fixed slippage rate cannot reflect how impact varies across stocks,…
This forum post presents a DAI query for filtering Chinese A-share records and asks how to save a large result table as an XLS spreadsheet for separate analysis. Its SQL selects dates, instruments, and derived indicators for price-limit events, including…
This Chinese research summary examines the turn-of-month calendar effect, described as stronger returns in the first half of a month than in the second. It reports that a simple strategy of buying during the first half and shorting during the second…
This research summary examines whether inventory cycle phases can help time broad sectors and the Shanghai Composite. It describes four phases—active restocking, passive restocking, active destocking, and passive destocking—and compares qualitative phase…
The document raises a data-timing question about China’s Dragon-Tiger List, a report of stocks with notable trading activity. It names fields for returns one, two, and five days after a stock appears on the list, and says the five-day return is unavailable…