OPEN-SOURCE SCRIPT
TE Feature Screen
Stop Trading on Lagging Indicators. Start Measuring True Predictive Power.
Are your indicators actually predicting the next move, or are they just taking credit for a trend that’s already happening?
Introducing the Transfer Entropy Feature Screen. Built for quantitative researchers and serious algorithmic traders, this script leverages Information Theory to mathematically separate predictive signals from random market noise.
Instead of relying on standard correlation, this engine uses Conditioned Transfer Entropy (TE). It deliberately strips out the stock's past momentum, forcing every indicator to prove it is bringing new, independent information to the table about where the price will be h bars from now.
Tailored Specifically for the Stock Market
This Equities Edition focuses strictly on institutional footprints and structural market mechanics:
Order Flow & Volume Kinetics: Track Limit Order Absorption (churn), Close Location Value (CLV), and Unusual Volume (RVOL) Z-scores to spot institutional accumulation before the breakout.
Structural Edges: Measure Overnight Gap extent and Standard Deviation Distance from the daily VWAP.
Index Relative Strength: Dynamically compare the stock's acceleration against a benchmark index (like the SPY or QQQ) to find true market leaders.
Built for Rigorous Quant Research
To prevent curve-fitting, the script features a strict Walk-Forward backtesting engine.
Input your exact calendar dates for In-Sample (Training) and Out-of-Sample (Testing) windows.
The engine benchmarks every feature against randomized "null shuffle" data to generate a definitive Z-score, proving whether an indicator's edge is mathematically significant or just a lucky streak.
Seamless Data Export: Bypassing TradingView's execution limits, the engine processes massive historical windows in the background and delivers a cleanly formatted CSV spreadsheet of the Z-scores directly to your email via alerts—perfect for building your own regime classification models.
Stop guessing which features matter. Let the math tell you what actually moves the market.
Setup Note for Users: To extract the data, set your IS/OOS calendar dates, toggle your desired features, and create a TradingView alert with "Send Email" checked. The script will automatically crunch the historical data and email you the CSV results upon the next bar close
Are your indicators actually predicting the next move, or are they just taking credit for a trend that’s already happening?
Introducing the Transfer Entropy Feature Screen. Built for quantitative researchers and serious algorithmic traders, this script leverages Information Theory to mathematically separate predictive signals from random market noise.
Instead of relying on standard correlation, this engine uses Conditioned Transfer Entropy (TE). It deliberately strips out the stock's past momentum, forcing every indicator to prove it is bringing new, independent information to the table about where the price will be h bars from now.
Tailored Specifically for the Stock Market
This Equities Edition focuses strictly on institutional footprints and structural market mechanics:
Order Flow & Volume Kinetics: Track Limit Order Absorption (churn), Close Location Value (CLV), and Unusual Volume (RVOL) Z-scores to spot institutional accumulation before the breakout.
Structural Edges: Measure Overnight Gap extent and Standard Deviation Distance from the daily VWAP.
Index Relative Strength: Dynamically compare the stock's acceleration against a benchmark index (like the SPY or QQQ) to find true market leaders.
Built for Rigorous Quant Research
To prevent curve-fitting, the script features a strict Walk-Forward backtesting engine.
Input your exact calendar dates for In-Sample (Training) and Out-of-Sample (Testing) windows.
The engine benchmarks every feature against randomized "null shuffle" data to generate a definitive Z-score, proving whether an indicator's edge is mathematically significant or just a lucky streak.
Seamless Data Export: Bypassing TradingView's execution limits, the engine processes massive historical windows in the background and delivers a cleanly formatted CSV spreadsheet of the Z-scores directly to your email via alerts—perfect for building your own regime classification models.
Stop guessing which features matter. Let the math tell you what actually moves the market.
Setup Note for Users: To extract the data, set your IS/OOS calendar dates, toggle your desired features, and create a TradingView alert with "Send Email" checked. The script will automatically crunch the historical data and email you the CSV results upon the next bar close
開源腳本
秉持TradingView一貫精神，這個腳本的創作者將其設為開源，以便交易者檢視並驗證其功能。向作者致敬！您可以免費使用此腳本，但請注意，重新發佈代碼需遵守我們的社群規範。
免責聲明
這些資訊和出版物並非旨在提供，也不構成TradingView提供或認可的任何形式的財務、投資、交易或其他類型的建議或推薦。請閱讀使用條款以了解更多資訊。
免責聲明
這些資訊和出版物並非旨在提供，也不構成TradingView提供或認可的任何形式的財務、投資、交易或其他類型的建議或推薦。請閱讀使用條款以了解更多資訊。