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.
16,761 documents
The article examines why A-share strategy results in historical tests can differ from live or simulated trading. It groups causes into data problems such as survivorship bias and price adjustments, model problems such as overfitting and crowded factors,…
A short forum post raises a practical data choice for stock strategies: whether to use prices adjusted for corporate actions when calculating indicators and position profit and loss. The author suggests that adjusted series may be suitable for indicator…
This tutorial builds a multi-index rotation strategy using nine Chinese equity index ETFs as proxies for different market sectors. It ranks each index by its return from twenty sessions earlier to the latest close, then selects the strongest index if that…
This guide explains how to run FMZ strategies through a local Python backtest engine. It describes passing settings such as dates, bar interval, exchange, instrument, and starting balances through a configuration string, then initializing the engine and…
The document explains quantitative trading software as a system for turning investment rules into data analysis, strategy development, historical simulation, risk controls, and automated orders. It describes typical inputs such as prices, volumes, and…
This research summary compares periodic rebalancing with buy-and-hold for portfolios combining stocks and bonds. It reports that monthly rebalancing can lag buy-and-hold when stocks trend strongly relative to bonds, because rebalancing sells some of the…
This report summary outlines an investigation of how high-frequency factors behave across different periods and domains, along with an analysis of what may influence their performance. The authors say earlier work built factors from trading logic using…
Dual Thrust is a breakout method that sets daily entry thresholds from recent price extremes. It calculates the highest high and close, and the lowest low and close over a lookback window, then takes the larger of two cross-range measurements. Multiplying…
This document describes a script for analyzing the structure of a strategy’s drawdowns from daily profit-and-loss data. It rebuilds an equity curve from a starting capital value, derives an underwater curve, and separates that curve into individual drawdown…
This educational MetaTrader 5 strategy trades continuation after a qualifying gap between the prior Nikkei 225 cash-session close and the new session open. It builds an opening range from one-minute bars, calculates session VWAP, and considers a long after…
The document presents a Composite Momentum oscillator intended to identify stocks with rising recent momentum. Its calculation combines changes in a short-versus-long moving average momentum measure with a recent high-low range oscillator, then smooths the…
This documentation describes Freqtrade’s plotting commands for viewing price candles, volume, strategy indicators, and trades from a database or backtest export. The dataframe plot can show price-scale indicators such as moving averages alongside separate…
This stock-selection rule combines three filters: rank stocks by relative volume and keep the highest-ranked group, require positive net institutional flow as reported in a trading disclosure dataset, and exclude stocks whose opening price is more than six…
This Chinese equity screening idea combines trading activity, price movement, and a measure of large-order net flow. It selects stocks with turnover between 3% and 12%, requires the product of the day’s percentage change and super-large-order net volume to…
This overview presents active quantitative management as a process of forecasting factor returns and managing portfolio exposures. It describes a linear model in which each stock’s return is associated with its factor exposures, factor returns, and a…
This forum exchange explains how to use a platform’s precomputed limit-up or limit-down factor in a stock backtest. The motivating problem is that adjusted closing prices cannot be directly compared with actual daily price limits, which are quoted in…
This stock-screening proposal combines three filters: today’s increase in institutional holdings must exceed five percent, the share price must have fallen for seven consecutive sessions, and the company’s 2019 dividend payout ratio must be above twenty-five…
This utility downloads historical quote data for a selected timeframe across every symbol in the platform’s Market Watch. Its purpose is to make data available for backtesting across multiple instruments while keeping the timeframe consistent. It can also…
The document describes an automated trading approach that opens an initial position in the direction of the current candle. When the trade moves favorably by a configured point threshold, the system adds volume to the position. It can begin trading once the…
This Python script launches a Rust grid-trading backtest for each symbol in a ticker configuration. It assembles daily market-data and latency-file paths for a specified date range, passes instrument and strategy settings to the backtest executable, and runs…
A brief Chinese-language forum exchange asks whether a futures main continuous contract has a standard contract code or whether users must combine the data themselves. The reply says live trading gateways do not provide such a continuous contract, while some…
This short indicator description presents an extended version of normalized RSI. It says the normalization approach is explained in a separate discussion, while this version adds options intended to make the indicator easier to interpret for trading…
This research report examines conditional generative adversarial networks (cGANs) for forecasting asset returns and covariance matrices, then using those forecasts in portfolio allocation. The model conditions on recent multi-asset returns and random inputs…
The document describes an Expert Advisor that trades from changes in the NonLagDot indicator’s color. A signal is acted on when a bar closes and the indicator changes color, so the rule is based on a confirmed bar rather than an intrabar change. The advisor…