Rolling Average Pvalue
Quick Reference
| Property | Value |
|---|---|
| Dimension | quality |
| Category | statistics |
| Version | v1.0 |
| Output Column | avg_pvalue |
Rolling average of a binary indicator 鈥?estimates confidence/probability over time
Formula
rolling_mean(indicator, window)
CDM Inputs
| Column | CDM Table | Description |
|---|---|---|
indicator | cdm_* | CDM source table |
Parameters
| Parameter | Type | Default | Description |
|---|---|---|---|
window | integer | 60000 | Window for rolling average |
Output
Column: avg_pvalue
Rolling mean of the binary indicator
Market Intuition & Trading Rationale
Rolling average p-value estimates the probability or confidence of a binary indicator over time: rolling_mean(indicator, window). The indicator could be anything 鈥?a statistical test result (is the signal significant?), a regime flag (is the market trending?), or a quality check (did the signal predict correctly?). The rolling mean converts a noisy binary sequence into a smooth probability estimate.
This is the simplest form of sequential probability estimation. A value of 0.8 means the indicator was true 80% of the time over the window. A value of 0.5 means the indicator is effectively random. The window controls the bias-variance tradeoff: short windows react quickly but are noisy; long windows are stable but slow to reflect changes.
Usage Cases
- Statistical significance tracking: Feed the result of a hypothesis test (e.g., "is IC significantly > 0?") as the indicator. rolling_average_pvalue tells you what fraction of recent periods showed statistically significant predictive power.
- Regime probability estimation: Feed a binary regime classifier as the indicator. The rolling average becomes the probability of being in that regime, smoothly transitioning as the market changes.
- quality context: Used in
jump_microstructure_noisepack 鈥?the indicator tracks whether a detected jump was statistically significant, and the rolling average estimates the jump detection reliability.
YAML Definition
name: rolling_average_pvalue
description: Rolling average of a binary indicator 鈥?estimates confidence/probability
over time
category: statistics
dimension: quality
version: v0.9.0 (Beta)
required_inputs:
- indicator
output_column: avg_pvalue
output_description: Rolling mean of the binary indicator
tags:
- quality
- confidence
- statistics
parameters:
window:
type: integer
description: Window for rolling average
required: false
default: 60000
formula: rolling_mean(indicator, window)