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.
21,023 documents
This code excerpt implements neural-network components for a momentum forecasting model based on a temporal fusion transformer design. It includes feed-forward layers, gated linear units, gated residual networks with skip connections and normalization, and…
This Chinese-language post describes a stock screen combining daily price range, positive institutional net flow, and a relative-volume band. The stated rules select stocks with an amplitude above one, institutional flow above zero, and volume ratio between…
A user reports that minute-bar open and close prices generated from CTP market data sometimes differ from values shown in mainstream trading software, with discrepancies of about one currency unit. The reply identifies a timestamp convention as a possible…
This forum discussion documents installation problems while setting up vnpy_ctp on a freshly reinstalled Intel Mac. The initial failure occurred during an editable package installation because pip could not obtain the required meson-python dependency. A…
The document describes a lag-reduction adjustment applied to a simple moving average. It first calculates a moving average over a configurable period, then uses the ratio between its current and prior values, raised through a logarithmic and exponential…
The article reviews the progression from earlier GPT language models to GPT-3, emphasizing scale and the ability to perform varied tasks from prompts or examples without task-specific fine-tuning. It describes code generation alongside broader language…
The document explains an oscillator that tracks one lagged-return autocorrelation estimate through time. At each bar, it calculates Pearson correlation between returns in a rolling window and the same returns shifted by a configurable lag. Positive readings…
This article presents a broad framework for choosing and assessing quantitative trading strategies. It recommends matching a strategy to an investor’s temperament, available time, preferred trading frequency, capital, programming skills, and goals. It argues…
This document describes a chart indicator that plots price levels associated with the Special Theory of Price Discovery. Users select a reference period—daily, weekly, monthly, quarterly, or yearly—and can adjust the price expansion multiples used to place…
The article defines a bounce number as the count of price touches at the two boundaries of a range before price reaches a take-profit level. It proposes measuring these counts across a symbol’s historical price data to inform lot sizing or the number of…
This article explains how to assess a systematic futures model through test design and several complementary performance measures. It recommends testing across multiple instruments and long historical periods, and including commissions and slippage. The…
This document summarizes two external research papers. The first concerns allocating capital among private fund strategies within a fund-of-funds portfolio, using multi-objective linear programming to derive allocation weights. The second examines the ratio…
The post describes a daily stock screen for main board shares with turnover between 3% and 12%, a daily gain above 1%, and positive return. It frames the turnover band as a way to focus on trading activity and the gain and return filters as profitability…
This short explanation describes a variation on the stochastic oscillator, a momentum measure that compares price with its range over a chosen lookback period. It notes that the oscillator’s responsiveness can be adjusted by changing that period or by…
The document describes why quantitative investors use machine learning to analyze large datasets, forecast market direction, and select portfolios. It highlights the ability of these methods to represent nonlinear relationships and process data efficiently,…
This indicator displays a table of moving-average values across earlier bars, along with sequential value changes and distances in points. Its inputs allow users to compare moving averages with one another or with bar prices such as open, high, low, and…
The indicator measures how far the current close sits above or below the intraday volume-weighted average price, expressed in units of a calculated standard deviation. VWAP resets each trading day. Once a minimum number of bars has elapsed, the…
This document introduces John Ehlers’ Voss Predictive Filter, an indicator intended to identify cyclical turning points with signals that may appear earlier than those from other indicators. Its calculation first applies a band-pass filter to price changes,…
This note describes a stock screen requiring daily price amplitude above 1, MACD above its zero line, and a positive price-to-earnings ratio. The conditions aim to combine elevated price movement, a bullish technical signal, and a company reporting positive…
This document describes a small MQL5 benchmarking library for comparing how quickly functions or expressions execute. It introduces three measurement styles: timing a call in microseconds per invocation, running a fixed number of repetitions and reporting…
This tutorial explains how to use Alphalens to examine whether factor scores are associated with future asset returns. It distinguishes factor research from portfolio backtesting: factor analysis helps characterize predictive power, consistency across…
Risk parity allocates portfolio weights so that assets contribute more evenly to total portfolio risk, rather than assigning capital according to expected returns or equal dollar amounts. The document outlines a workflow: estimate asset volatilities and…
The document presents a Chinese equity screening rule that selects stocks with turnover between 3% and 12%, circulating market value between 5 billion and 10 billion yuan, and a flagged rounded or arc-shaped price pattern. It supplies equivalent screening…
The report argues that ranking fund managers can reduce redundant information when one manager oversees several similar funds. It proposes tracking manager style across market capitalization, value versus growth, industry concentration, individual-stock…