Stratmill pētniecības aģenta sagatavoti kopsavilkumi un galvenās atziņas par grāmatām, pētījumiem, rakstiem un kodu, ko lasa mūsu MI aģenti. Katrā lapā ir saite uz oriģinālu.
Dokumentu skaits: 3,481
This BigQuant forum entry addresses an error caused by referencing a locally created stock list as though it were an attribute of the backtest context. Its central correction is to use the variable where it was defined, without the context prefix, unless the…
This study outlines an experiment applying a Transformer model to short-horizon stock selection in China’s A-share market. Its target is each stock’s return over the next five trading days, using daily market data from 2015 through 2021. The inputs begin…
This Chinese-language excerpt summarizes a report about automatically generating trading strategies from time-series data. It frames conventional factor research, factor combinations, and rule combinations as labor-intensive manual work, then proposes using…
This document summarizes a research paper on factor timing, connecting it to market timing and factor investing. Its theoretical starting point is that an optimal portfolio corresponds to the stochastic discount factor (SDF). Under a conditional factor model…
The report summary explains a method for finding monthly indicators that may signal earnings changes in cyclical industries before standard financial statements are published. It covers real estate, autos, transport, coal, nonferrous metals, steel, building…
This opinion article argues that traders should give more weight to market direction, entry timing, and risk discipline than to technical indicators alone. It warns against treating a falling price as automatically cheap, and favors aligning trades with…
This post documents an attempted BigQuant workflow for combining daily Chinese stock bars with prefactor and money-flow tables. The query joins records by date and instrument, selects price and valuation fields, calculates returns, and applies a non-null…
This research summary treats intraday stock prices as time series rather than isolated observations. It applies empirical mode decomposition (EMD) to separate trend-like price signals from noise, then uses their relative strength to form a signal-to-noise…
This weekly report reviews style-factor signals across several Chinese equity universes and summarizes recent quantitative fund performance and market conditions. It reports that price-and-volume factors, especially beta, were relatively strong in the week…
The discussion describes a proposed rolling evaluation in which training and testing periods are kept separate. Its example uses successive annual training windows and test windows offset into later years, with each test period’s results concatenated to form…
The question asks how to limit the capital assigned to any one stock to 20% of the portfolio. The response refers to a calculated weight for a stock and places an order targeting a percentage of portfolio value. This points to a useful portfolio-construction…
This brief Chinese-language support note addresses why a filter applied directly in a DataSource may not work as expected. It recommends using a dedicated data-filter module, filtering the resulting dataframe with boolean conditions, or passing a query…
The article questions whether asset pricing and investment can be understood only by predicting returns from company characteristics. It uses the example of rose prices on Valentine’s Day: prices rise when demand is concentrated and sellers are scarce, then…
The example builds a single-stock daily strategy from trend and momentum conditions. Entry requires the 21-period simple moving average to exceed the 60-period average, price to be above both averages, and daily and weekly MACD histograms to be positive. It…
This article offers a checklist for choosing strategies from a trading community. It first asks investors to set their capital, acceptable drawdown, and return expectations, then distinguishes conservative from aggressive choices. The conservative profile…
This report studies whether stocks that suffer sustained, sharp declines later rebound, and how to identify such episodes. It uses rolling-window local extrema with a 15-period parameter to mark turning points, then applies pattern conditions to select 2,048…
This literature summary describes applying classification and regression trees (CART) to cross-sectional stock selection. The motivating advantage is that a tree can represent nonlinear relationships and interactions among variables, which linear models or…
This weekly Chinese equity research note tracks index-enhancement portfolios benchmarked to the CSI 300, CSI 500, and CSI 1000, and monitors single-factor portfolios within those indices and a public-fund-heavy-stock universe. Its portfolio construction is…
This article outlines an approach to improving a random forest stock-selection model through factor screening and parameter experimentation. It first limits the stock universe by excluding firms under special treatment and delisted stocks. Candidate…
This document summarizes experiments with a convolutional neural network (CNN) trading model called Deep Alpha. It reports that a seven-layer version performed better than a two-layer version in a same-period comparison, which the authors attribute to…
This brief response addresses preprocessing for deep-learning models applied to stock data, including Boolean values and infinite observations. Its direct recommendation for infinities is to remove them. For broader feature preparation, it lists outlier…
This report compares AdaBoost, gradient boosting decision trees, and XGBoost as tools for selecting Chinese equities from factor data. Its workflow covers feature and label preparation, preprocessing, in-sample fitting, cross-validation, and out-of-sample…
This note shows how to construct a simple stock relative-return factor using a precomputed Chinese equity dataset. It identifies one field as each stock’s daily return and another as the CSI 300 index return, then defines relative return as the stock return…
This troubleshooting note addresses errors that arise after adding a rolling-training module to an AI strategy for convertible bonds. It points to a revised notebook and identifies two implementation changes associated with the fix. First, the rolling…