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.
79,386 documents
This brief indicator note describes a timeframe-selection option for ColorStepXCCX. The user can choose the chart period on which the indicator operates through an input setting; the listed default is four hours. The note also identifies the separate…
This MetaTrader indicator plots a price channel, trend line, Fibonacci levels, and Fibonacci fan using recent peaks identified by an NRTR-based ZigZag indicator. Users can select a calculation timeframe and adjust the ZigZag period, display options, colors,…
This Chinese-language post describes a stock selection screen using three conditions: turnover between 3% and 12%, appearance on the market’s trading leaderboard the previous day, and circulating share count no greater than 5.5 billion. It frames the…
The document outlines a financial time-series study using returns on the CSI 300 index. It says the analysis applies TGARCH and related derivative models, compares their behavior, and attempts to fit and forecast both prices and volatility. This frames…
This post describes a Chinese equity screen that combines three conditions: daily price range greater than one percent, a trading date in 2021, and a nonempty name for an outstanding convertible bond. It provides example implementations in a charting formula…
The document explains the behavioral challenges of trend following: traders must tolerate repeated small losses, fluctuations in open profits, and periods of waiting for a large move. It recommends evaluating a strategy over many trades rather than judging…
This podcast recap describes Loop Crypto’s approach to recurring cryptocurrency payments. Because ordinary crypto wallets send funds through user-initiated transactions, collecting subscriptions is less convenient than card-based payment collection. Loop’s…
This Chinese stock-selection post combines three screening ideas: rank stocks by turnover or “fund strength,” rank them by net large-order volume, and require positive but capped gains over the recent ten trading days. It describes turnover as a proxy for…
This document explains a swing indicator that anchors an Archimedean spiral at the last confirmed pivot. Its structure changes when price makes a new high or low over a specified lookback; when direction flips, it places the pivot retroactively at the…
This document presents a Chinese stock screening rule that combines a five-period amplitude threshold, a change in a proxy for institutional trading activity, and a newly formed bullish KDJ crossover. The crossover condition requires the J line to move above…
This document explains how to extend a DSS Bressert semaphore indicator with sound alerts, email messages, and mobile push notifications for buy and sell signals. It describes configurable inputs for selecting the signal bar, enabling each notification…
This document describes an automated price action strategy that compares highs and lows across multiple chart timeframes, then looks for lower-timeframe candle closes that break relevant structure. It illustrates the approach with comparisons among hourly,…
This Chinese stock-selection rule combines turnover between 3% and 12% with a rising DEA condition and a KDJ stochastic K reading from 0 to 20. The post describes the low K reading as a possible trough area and treats rising DEA as a sign of improving trend.…
This Chinese equity screen selects main-board stocks linked to the robotics concept, with turnover between 3% and 12%, floating market value below 10 billion yuan, and prior-day trading value above 60 million yuan. It combines a theme classification with…
The document describes a MetaTrader indicator that draws daily high-low boxes for the Asia, London, and New York trading sessions. It uses hourly data, adjusts broker timestamps to GMT with a configurable offset, and supports sessions that cross midnight.…
This stock-selection proposal screens Chinese equities using three conditions: RSI below 65, circulating market capitalization between 5 billion and 10 billion yuan, and turnover between 2% and 9%. It presents the screen as a way to combine a technical…
This support note explains a model-training failure attributed to problems in the input data. The specific diagnosis is that an erroneous expression caused the buy-condition factor to fail extraction. It also notes that the dataset had only two factors, both…
This document describes a trend indicator implemented using the NRTR approach, with its calculation timeframe set through an input parameter. Average True Range is used in constructing the indicator’s moving average, so the method incorporates recent price…
The document introduces the Sharpe ratio as a measure of excess return per unit of portfolio risk, then describes a method for forecasting how that ratio changes over time. It outlines an approach associated with Whitelaw that estimates Sharpe ratios from…
This Chinese-language post proposes a short-term stock selection screen using daily amplitude, a measure described as today’s controlling-shareholder change, and a ratio combining yesterday’s turnover with today’s auction volume relative to yesterday’s…
This guide explains what happens when an OctoBot cloud strategy is stopped and how to restart it. Stopping cancels the bot’s outstanding buy and sell orders and prevents it from placing new orders. If the corresponding option is selected, the bot also sells…
The document surveys a library of trailing-stop methods intended to manage open positions independently of entry rules. Methods include trailing behind recent bar highs or lows, fractal extrema, stepped price advances, tighter distances as profit grows,…
This proposed screen selects main-board stocks using an RSI ceiling, a positive daily gain, and a comparison involving the prior session's opening auction price. Eligible stocks are ranked by auction amount, with the highest-ranked names selected for a small…
This BigQuant example builds a binary classifier from rolling windows of daily open, high, low, close, and volume changes. Its network applies a convolutional layer to each 50-step, five-feature input, passes the result through an LSTM and dense layers with…