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 Python class implements a broad collection of formula-based equity signals using daily close, open, high, low, volume, returns, and volume-weighted average price data. Its methods translate rank, correlation, rolling-window, change, volatility, and…
This guide explains how to use VeighNa’s CTA strategy module to load strategy classes, create instances, initialize them with historical data, and start or stop automated trading. It describes how instances can trade different futures contracts with separate…
This short note highlights that feature importance can differ across predictive models, naming StockRanker and XGBoost as examples. It also points to changing the number of leaf nodes and the learning rate as model-tuning choices. These observations can help…
This brief options-forum exchange asks how an options strategy should handle a position that reaches expiration during a backtest. A respondent says to calculate the profit or loss using the closing price on the final trading day. The answer offers a simple…
This report describes a high-frequency futures CTA approach that uses the relationship between traded prices and adjusted intraday open interest. It argues that open interest can provide information beyond trading volume, and derives directional signals from…
This stock screen combines a turnover-rate range of 3% to 12%, exclusion of Beijing-listed A shares, and a rounded price-pattern condition. The article frames turnover as a measure of capital activity and the chart pattern as a possible indicator of upside…
This article describes a factor test of price-to-earnings ratio as a stock selection signal among CSI 300 constituents. It divides stocks into five groups, then compares group returns under weekly and monthly rebalancing. The reported qualitative result is…
The document describes a trade-history analysis that rebuilds trades from historical deals and reports metrics intended to assess entry and exit efficiency. Maximum adverse excursion (MAE) measures the largest move against a position while it is open, while…
The document addresses a timing problem when using weekly indicators in a daily trading strategy. A weekly calculation may be absent or incomplete during the current week if the data pipeline only forms values after the week closes. Its example asks how to…
MarketPredictor is described as an open-source MetaTrader 5 Expert Advisor that applies mathematical tools to historical prices and future scenarios. Its stated methods include Fast Fourier Transform analysis to identify dominant frequency patterns and Monte…
The document outlines a basic backtesting workflow: define a strategy, gather historical data, calculate returns and supporting statistics, then decide whether to deploy or refine the idea. It notes that crypto strategies may use pairs trades, rebalancing…
The post describes a stock screen combining three conditions: turnover between 3% and 12%, a rising DEA signal, and positive afternoon large-order flow. It frames the filters as measures of trading activity, short-term direction, and buying pressure. The…
A platform user reports that a backtest transaction record can misstate realized profit and loss when a stock position passes through an ex-rights corporate action. The example compares transactions in Chinese equities before and after such an adjustment and…
The post outlines an intended Chinese equity strategy that buys stocks when their price-to-earnings ratio ranks near the low end of its rolling history and sells when it reaches the high end, while restricting the universe to companies above a large…
The guide explains how to run a strategy or crypto basket in paper trading through OctoBot cloud. A user starts trading from the selected strategy or basket, chooses paper trading, and sets the size of a virtual portfolio. The simulated bot follows the…
The Expert Advisor combines an initial buy or sell position with a protective stop order, then builds a grid of pending orders at set price intervals. It uses ADX on a configurable timeframe to choose the grid style: below 40 it places limit orders, while…
This FMZ Pine Script reference describes script structure, declarations, indentation, comments, and time-series behavior. It explains how historical values are accessed, illustrates calculations over bars, and discusses limits on historical references. It…
This article explains a GBDT-plus-logistic-regression method for classifying stock direction. Gradient-boosted decision trees capture nonlinear feature relationships; each tree maps observations to leaf nodes, which are encoded as indicator variables.…
This Chinese-language community post outlines a stock screening rule combining three technical conditions: daily amplitude above one percent, a newly formed KDJ golden cross, and three consecutive sessions in which the close is below the open. It frames high…
This research note describes a tactical asset allocation method that converts the tone of company and macroeconomic news into a slower-moving signal. News sentiment scores are aggregated weekly, then filtered with a two-sided cumulative sum (CUSUM) procedure…
This futures strategy tracks the crush spread formed from soybean, soybean meal, and soybean oil contracts. It weights the three contract prices using an approximate processing ratio, computes the spread as the combined meal and oil value less the soybean…
This report presents machine learning as a data-driven approach and distinguishes supervised learning, which maps labeled inputs to outputs, from unsupervised learning, which seeks structure in unlabeled data, and reinforcement learning, which learns actions…
The document describes a short-term stock screen combining three technical conditions: price range greater than one, a gain below six percent at the 9:25 observation, and three consecutive bearish daily candles. It frames consecutive down days as a possible…
This report introduces machine learning as data-driven learning and outlines its three broad forms: supervised, unsupervised, and reinforcement learning. It describes how computing advances, larger datasets, and accessible software libraries have helped make…