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
This platform guide explains how FMZ defines live trading: a strategy instance connected either to an exchange’s production environment or to its simulation environment. Starting an instance requires a saved strategy, a deployed platform host, and at least…
The code implements a Cox–Ingersoll–Ross model for a mean-reverting portfolio, extending an Ornstein–Uhlenbeck model interface. It supports fitting the model to portfolio prices or to two assets, and estimates the long-run mean, reversion speed, and variance…
This article develops two monthly equity factors from Chinese market flow data. The first, called institutional-versus-retail harvesting, adds the cross-sectional ranks of large-order net buying and small-order net buying. The author interprets the factor as…
The post answers why a community-edition backtest can work while simulated trading produces no orders. It distinguishes imported historical data, which can be replayed during a backtest and used to initialize indicators, from simulated trading, where…
The discussion explains why a strategy can run in a historical backtest yet place no trades in simulated trading. Imported historical bars support backtests and indicator initialization, while a live simulation generally depends on real-time tick or bar…
The document describes a MetaTrader 5 indicator intended to mark buy and sell points around false breakouts of support and resistance, while also displaying dynamic levels and potential market tops and bottoms. Its stated logic identifies a top when a…
This article explains how to turn future returns into labels for supervised trading models, arguing that the label should match the instrument, trading horizon, liquidity, and costs of the intended strategy. It first describes using an n-period return as a…
This summary of a securities research report examines alternative data, focusing on company job postings. It describes a hiring-count factor tested over 2017–2021, reporting that preprocessing and filtering job listings improved its backtest performance. The…
This brief forum exchange explains how VeighNa handles strategy updates after the application has started. For backtesting, users can use the strategy reload control to rescan strategies. For a live trading instance, the reply says that changing instance…
The article compares two Chinese equity approaches built around market capitalization. The traditional strategy selects the smallest-cap stocks for a portfolio. The AI version uses a StockRanker ranking model with market capitalization as its sole feature.…
The document describes a simple Chinese equity screening idea: select stocks with an amplitude above a stated threshold, a positive weekly MACD histogram, and membership in the metaverse concept group. It presents the criteria as a way to combine a…
This proposed equity screen combines RSI below 65, a product of percentage price change and a large-order net-flow measure above 1, and a sequence of three declining sessions. The accompanying explanation says the RSI filter supplies technical context, the…
The document describes a technical stock screen using a range threshold above 1, RSI below 65, and MACD above its zero line. It interprets the range as a sign of greater price movement, the RSI ceiling as avoiding an extremely overbought reading, and…
This excerpt introduces a framework for assessing whether stock-selection factors may deliver persistent returns beyond broad market exposure. It distinguishes market-related beta from alpha and presents factor effectiveness as a question of both return…
This implementation models a spread as a collection of instrument legs, with separate multipliers for calculating its quoted price and translating spread quantities into leg quantities. It combines leg bid and ask prices, reversing which side is used for…
The document is a historical series of hourly candlestick observations for a BCH/USDT market. Each row records a timestamp, open, high, low, close, and traded volume, giving the basic inputs commonly used to inspect price movement, calculate technical…
The document contains historical ADA/USDT candlestick observations at half-hour intervals. Each row records a timestamp, open, high, low, close, and traded volume, allowing a researcher to inspect price movement and activity or use the series as an input to…
This strategy uses a fast and a slow moving average to trade a cryptocurrency futures contract. A bullish crossover opens a long position or reverses a short position; a bearish crossover opens short or reverses long. Signals use earlier completed bars…
The document shows a simple workflow for evaluating two futures strategies together. It runs separate historical simulations for an ATR-RSI strategy on an equity index contract and a Bollinger channel strategy on a metal contract. Each run specifies its own…
This document explains how to build a multi-contract strategy using synchronized bar data, per-leg targets, and order management. Its example computes the spread between two weighted contract prices, updates a rolling window, and uses Bollinger Bands to…
The document describes a two-leg spread strategy built around Bollinger Bands. It calculates a weighted price difference between two contracts, samples the spread on a five-minute schedule, and compares it with a rolling mean and standard deviation. A move…
This document is a daily candlestick dataset for the BTC/USDT market during 2019. Each row records a timestamp and the open, high, low, and close prices, together with traded volume. The visible entries span portions of the year, including early-year…
The document contains hourly open, high, low, close, and volume observations for the ADA-USDT market. The visible records begin in early May 2018 and continue through the end of December 2018, with gaps in the displayed sequence. The fields support basic…
This document explains the structure of a historical trading strategy backtester. It loads bar or tick records over a selected date range, initializes a strategy with a warm-up period, then replays the remaining data. The engine tracks simulated orders and…