Linear Regression Slope

Quick Reference

Property	Value
Dimension	regime
Category	statistics
Version	v1.0
Output Column	trend_slope

Linear regression slope over a rolling window 鈥?measures trend direction and magnitude

Formula

linear_regression_slope(value, window)

CDM Inputs

Column	CDM Table	Description
value	cdm_*	CDM source table

Parameters

Parameter	Type	Default	Description
window	integer	3600000	Window for regression estimation

Output

Column: trend_slope

Slope coefficient from rolling linear regression

Market Intuition & Trading Rationale

Linear regression slope fits a trend line over a rolling window: linear_regression_slope(value, window). Positive slope = uptrend (value is rising over time). Negative slope = downtrend. Near-zero slope = sideways/ranging. The slope magnitude indicates trend strength 鈥?steeper slopes mean faster price change. This is a more robust trend measure than simple return_momentum because it uses all observations in the window, not just endpoints.

Usage Cases

• Trend direction and strength: Positive + large = strong uptrend (go long). Negative + large = strong downtrend (go short). Near zero = ranging (avoid directional strategies).
• Trend exhaustion: Slope that was large and is declining 鈫?trend is losing momentum. Tighten stops, prepare for reversal or consolidation.
• context: Used in order_flow_accumulation pack 鈥?slope of cumulative OFI trend confirms whether accumulation is translating into directional price movement.

YAML Definition

name: linear_regression_slope
description: Linear regression slope over a rolling window 鈥?measures trend direction
  and magnitude
category: statistics
dimension: regime
version: v0.9.0 (Beta)
required_inputs:
  - value
output_column: trend_slope
output_description: Slope coefficient from rolling linear regression
tags:
  - regime
  - trend
  - statistics
parameters:
  window:
    type: integer
    description: Window for regression estimation
    required: false
    default: 3600000
formula: linear_regression_slope(value, window)