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 page introduces a lesson on using correlation coefficients to identify stock candidates for pairs trading and then building a strategy to trade them. It frames correlation as a screening tool for finding potentially suitable arbitrage pairs, followed by…
This document presents a Chinese equity screening rule that selects stocks with daily turnover between 3% and 12%, parent-company net profit growth between greater than 20% and up to 100%, and tradable shares capped at 5.5 billion. It describes the…
This documentation explains how Zipline organizes risk and performance measurements for algorithm simulations. A metrics set defines which values a backtest tracks, and its metrics can report at different frequencies. The default set includes examples such…
Adaptive Volatility Analysis (AVA) is an indicator intended to measure whether market volatility is increasing or easing. It compares a short-period exponential moving average of ATR with a longer-period exponential moving average, producing a relative…
This study compares Bitcoin returns during a defined US daytime window with returns outside that window, using hourly Gemini data from 2015 through 2024. It splits results around the October 2021 launch of the BITO fund as a rough proxy for Bitcoin’s growing…
The article contrasts quantitative trading, which it describes as using statistical models and systematic execution to exploit price deviations linked to investor emotion, with discretionary investing, where patience, independent judgment, and resilience may…
This A-share screening rule combines three conditions: RSI below 65, tradable market capitalization between 5 and 10 billion yuan, and a ten-day price gain above zero but below 35 percent. The accompanying Python example calculates a 14-period RSI and…
This A-share stock screen selects names whose prior trading range exceeds a stated threshold, whose opening price is within five percent of the ten-day moving average, and whose registered province is not Beijing. The document provides example formulas and…
This 2018 report revisits whether the investment clock can guide asset allocation amid trade tensions and political shocks. Its traditional mapping assigns equities to recovery, commodities to overheating, bonds to recession, and cash to stagflation. It…
This essay argues that systematic traders should separate the quality of a signal from the trader’s execution of it. A trader following a defined process should execute each signal consistently rather than deciding whether to take a trade based on recent…
This post describes a rule-based screen for Chinese equities. It selects stocks with turnover between 3% and 12%, circulating market value from 5 billion to 10 billion yuan, and circulating share capital no greater than 5.5 billion shares. The stated…
The Enhanced Oscillator Trigger (EOT) combines three normalized oscillators with smoothing and regression-based components. Each oscillator applies a high-pass filter and a supersmoother to price-derived data, then uses a fast-attack, slow-decay peak…
This brief indicator note describes a weighted deviation measure that applies linear weighting throughout its calculation. It distinguishes the measure from a built-in deviation function that uses a linear weighted moving average as its averaging method. The…
This note describes an indicator that represents financial asset price oscillations as a spectrum. Colored bands on the chart correspond to different oscillation harmonics, with their amplitudes displayed at the right side. The concept offers a visual way to…
The document explains the KDJ oscillator as a tool for identifying potential market entry conditions. It defines the raw stochastic value (RSV) as the close's position within the highest-to-lowest price range over a chosen period, scaled to 100. K and D are…
This indicator lets the user construct an input price series by combining the open, high, low, and close prices with four configurable coefficients, then applies a moving average to that series. The approach extends conventional moving-average inputs by…
The indicator described here anchors Fibonacci analysis to confirmed swing highs and lows rather than arbitrary chart points. It calculates retracement and extension references, then compares them across three configurable timeframes. When multiple levels…
This document describes a Chinese equity screen that selects stocks with at least five overlapping moving averages, an indicated 9:25 a.m. gain below 6%, and observations from 2021. The proposed rationale is that converging averages reflect alignment across…
The document raises a practical trading question: how to enforce a minimum holding period using trading days rather than calendar days. Its example calculates elapsed time by subtracting the last rebalance date from the current date, which counts weekends…
This brief description identifies SMI_Correct as a variation of the Stochastic Momentum Index, drawing on William Blau’s work on momentum, direction, and divergence. It attributes the indicator to an author and notes that the implementation uses reusable…
This document describes a version of the Commodity Channel Index smoothed with Tillson’s algorithm. Its defining behavior is recalculation on every incoming tick, with the indicator drawn in real time. The description also notes an implementation detail:…
The article explains an RSI divergence indicator that avoids confirmed pivots. Instead, it compares the current price and RSI with extremes from an earlier short window, then requires elevated volume and an RSI extreme before marking a bullish or bearish…
This article explains how the Fama-French model extends CAPM first with size and value factors, then with profitability and investment factors. It describes double sorts by size and book-to-market, return on equity, or asset growth, and forms RMW from…
The document explains how combinatorially symmetric cross-validation (CSCV) can estimate the probability of backtest overfitting (PBO). It distinguishes training overfitting, where a model performs poorly on held-out data, from backtest overfitting, where a…