Stratmill tyrimų agento parengtos knygų, straipsnių, mokslinių darbų ir kodo, kuriuos skaito mūsų DI agentai, santraukos ir pagrindinės mintys. Kiekviename puslapyje pateikiama nuoroda į originalą.
3,481 dokumentų
This article collects practical lessons from building supervised machine learning strategies for stock selection. It recommends keeping training and test data separate and splitting financial time series chronologically, since random splits can mix market…
This Chinese equity screening proposal combines three conditions: a price amplitude above 1%, free float no greater than 5.5 billion shares, and short-, medium-, and longer-term moving averages aligned upward. The article interprets the amplitude filter as…
This report examines whether updating fundamental signals more frequently can improve stock selection during China’s annual report season. It argues that factor signals decay over time, but more frequent rebalancing offers limited benefit for fundamental…
This Chinese-language research report outlines an equity index evaluation framework based on Barra-style risk factors. It proposes decomposing index returns into market, industry, and style contributions, including size, valuation, and volatility, then using…
This post presents a rules-based foreign exchange strategy intended to trade rebounds within a trend. It uses a short moving average channel derived from prior highs and lows, together with MACD sign as a directional filter. Long entries are placed as limit…
This study compares three futures signals: cumulative return over a lookback window, an average based on cumulative daily steps, and a trend measure incorporating Garman–Klass volatility. It reports information-coefficient statistics, groups contracts into…
This research summary describes a multi-factor tracking framework built around the pricing logic of factors. It examines historical returns for technical, expectation-based, and financial factors using a modified cross-sectional regression approach. To…
This excerpt summarizes a dated market note about the China 50 ETF and volatility conditions. It interprets a rise in the trading-volume put-call ratio as a shift toward cautious sentiment and notes a premium in at-the-money call implied volatility relative…
This article explains the All Weather approach as a way to build a portfolio without relying on precise economic-cycle forecasts. It frames the macro environment along growth and inflation dimensions, then describes matching stocks, bonds, commodities, gold,…
This BigQuant example addresses a limitation in which its high-frequency feature extraction module calculates within a single day, while a researcher may need features from minute data spanning multiple days. It shows a distributed workflow using FAI: read…
This overview explains how high-frequency trading can profit from price movements and fragmented U.S. securities markets. Because a listed security may trade on multiple exchanges, differences in liquidity, participants, or information timing can create…
The document surveys a platform’s patent database and describes how patent characteristics are turned into equity-selection factors. It groups patent information by patent type, application or grant status, and measures such as counts, claims, citations,…
The document explains how to set custom trading intervals in a trading engine by using a trading calendar. It points to two examples: rebalancing on Mondays and rebalancing at the start of each month. These schedules let a strategy follow calendar-based…
This post presents a China equity screening rule combining a turnover-rate band of 3% to 12%, outside volume greater than 1.3 times inside volume, and a rise of more than 5% in the stated close-position ratio. Its example formulas add a minimum volume…
The article compares decade-scale changes in six major equity indices and argues that long-run wealth accumulation depends on the balance between advances and declines, rather than on the count of up and down sessions. It describes a pattern of relatively…
This BigQuant strategy tutorial outlines a framework for combining a broad-market risk signal with individual-stock timing and position management. The proposed workflow builds separate feature expressions for index-level risk and stock-level conditions,…
The article compares a StockRanker stock-selection strategy with a Deepalpha deep neural network using the same 17 short-term factors, training data, and backtest setup. The strategy selects the top-ranked A-share stock each trading day and sells it the…
This weekly Chinese equity-market note describes a market-regime framework using the distance between 20-day and 120-day moving averages on a broad market index, alongside a measure of market-wide profit-making activity. In the reported week, the short…
The document introduces generative adversarial networks (GANs), a framework that trains two models against each other. A generator creates samples intended to resemble the training data, while a discriminator estimates whether a sample is real or generated.…
This summary of a research report examines the Chinese New Year calendar effect in mainland Chinese equities and how it relates to common quantitative factors. It reports that the broad market historically had an unusually favorable risk-adjusted profile…
The document explains how Level 2 equity market data extends a conventional five-level view to ten bid and ask levels, with aggregate depth, order queue details, and individual order records. It describes possible intraday interpretations of visible…
This study examines whether the number of executions recorded in tick data can improve a conventional reversal factor for Chinese A-share equities. It divides the traditional factor using trade-count information and proposes a refined reversal signal. The…
The post asks how to reproduce a Chinese stock industry return series from constituent-level data. It identifies three aggregation methods: equal weighting, weighting by total shares, and weighting by free-float shares. The example query retrieves…
The paper proposes a decision-focused learning framework for prediction followed by optimization when the downstream objective includes linear soft constraints. Such constraints penalize violations with a maximum function, which is nondifferentiable at its…