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
The document summarizes research on improving a primary sequence-learning task by selecting among related tasks that differ in timing. Examples include translation with different input delays and stock trend prediction over different future horizons. Because…
The document introduces linear regression as a way to describe how one variable changes in relation to another. It frames the model as a fitted straight line, with an intercept and slope, and uses daily returns of a Chinese bank stock and the CSI 300 as an…
This analysis proposes heterogeneous volatility as an alternative to the standard volatility factor in multi-factor equity selection. Standard volatility is the dispersion of a stock’s returns. The proposed measure instead captures the volatility of returns…
The article presents a discretionary stock setup organized around a 5-1-3 sequence. First, a share price is expected to build a base with at least five advancing sessions, allowing a minor interruption. Next, a large-volume bullish candle pushes toward the…
The document raises a label-design question for equity strategy development. It observes that labels based only on shifted closing and opening prices may not distinguish stocks that rise from those that fall, and asks whether adding highs, lows, and volume…
This research summary describes a stock-selection factor intended to measure herding in China’s A-share market. It applies the LSV model to intraday transaction records, classifying trades as buyer- or seller-initiated to estimate the relative strength of…
The document proposes a factor based on order-book imbalance and trading activity. It defines large orders using displayed volume across the first three bid and ask levels, identifies minute bars when buying volume exceeds selling volume, and weights returns…
The document sketches a thematic stock selection approach that seeks companies associated with currently popular market concepts. It suggests identifying themes through signals such as market sentiment, news, and social media, then investing in stocks linked…
The article offers three observational heuristics for retail traders trying to identify possible algorithmic activity in Chinese stocks. It points to jagged intraday price action and regularly sized, rapidly changing orders; similar trading amounts across…
The document reports a possible limitation in an A-share stock-screening field: setting the ST status value to zero did not keep a stock out of a copied strategy’s backtest after it moved from a special-treatment designation to a delisting-period name. The…
The document examines how information coefficients (ICs) are used to assess stock-selection factors and assign weights in multifactor models. It distinguishes Pearson correlation from rank correlation and points out a gap in the usual inference: a larger IC…
This research note examines why stocks with high correlation to their benchmark may perform better than low-correlation stocks in the China A-share market, despite findings cited for overseas markets. It explains benchmark correlation in terms of constituent…
This article proposes grouping a StockRanker model's scores instead of assuming that the highest-ranked stock is consistently the best choice. It notes that which rank performs well can vary from day to day, especially for models whose styles are unstable.…
This brief BigQuant support exchange explains a feature-name mismatch that prevents a LightGBM module from locating an input column. The module is given a calculated expression as the feature name, while the DataFrame column is actually named “ma.” The…
This document presents a rule for flagging Chinese stocks that rise sharply during the session and then give back a substantial portion of that move. Using minute closing prices, it tracks the running high, finds the lowest close before that high, and…
The document explains performance attribution as a way to identify where a portfolio’s returns and risks come from, assess whether excess returns may persist, investigate large fluctuations, compare strategies, and monitor the investment process. It outlines…
The document summarizes research that forecasts stock industry returns using lagged returns from a broad set of industries. Because including all industry predictors creates a high-dimensional regression with overfitting risk, the method uses LASSO to select…
The document explains how to read the balance sheet, income statement, and cash flow statement as connected views of a company’s resources, operating results, and cash movements. It distinguishes operating assets from investment assets, and describes how…
This research summary compares factor-based Smart Beta products with active quantitative and index-enhancement strategies. It reports that applying commonly used Chinese equity alpha factors to S&P 500 constituents produced little excess return before fees,…
This research summary presents a macro timing framework for Chinese equities and bonds built around monetary conditions and credit growth. It treats policy shifts as potentially informative because monetary policy is described as delayed and constrained by…
The document raises a portfolio allocation question: when a strategy’s performance weakens and later recovers, allocations based on its current Sharpe ratio may react too slowly. It asks whether changes in Sharpe ratio or the slope of a strategy’s cumulative…
This report describes using Campisi and Brinson models to analyze the performance and investment styles of bond funds, with fund selection for fund-of-funds managers as the motivation. It notes that bond fund valuation and disclosure challenges make…
This market-timing weekly report combines a broad-market regime filter, short-term risk assessment, industry views, valuation, and suggested equity exposure. It classifies the market using the distance between 20-day and 120-day moving averages of a broad…
This forum exchange addresses a BigQuant error reporting a missing “position” field. The response describes two changes: set the input mode of the feature-input module to SQL, and revise the weight variable used in the backtest function of the trading…