Intraday Reversal Probability
Quick Reference
| Property | Value |
|---|---|
| Dimension | signal |
| Category | mean_reversion |
| Version | v0.9.0 (Beta) |
| Output Column | reversal_score |
Intraday reversal probability score - estimates likelihood that recent directional movement will reverse
Formula
neg_autocorr(returns, window)
CDM Inputs
| Column | CDM Table | Description |
|---|---|---|
returns | cdm_* | CDM source table |
Parameters
| Parameter | Type | Default | Description |
|---|---|---|---|
window | integer [5, 1000] | 50 | Window for negative autocorrelation strength measurement |
Output
Column: reversal_score
Estimated probability of near-term price reversal (0-1)
Market Intuition & Trading Rationale
Intraday reversal probability estimates the likelihood of a near-term price reversal by measuring the negative autocorrelation of returns: neg_autocorr(returns, window). When returns exhibit strong negative autocorrelation (today's up move tends to be followed by tomorrow's down move), the reversal probability is high — the market is mean-reverting. When autocorrelation is positive (trending), reversal probability is low.
This feature captures a well-documented microstructure phenomenon: at short intraday horizons (seconds to minutes), returns tend to be negatively autocorrelated due to bid-ask bounce, inventory management by market makers, and liquidity-driven overreactions. A large buy order pushes the price above fair value; market makers then sell to rebalance, pushing the price back. Reversal probability quantifies how reliably this pattern occurs for the current instrument.
The window parameter determines the timescale of the reversal pattern. Short windows capture bid-ask bounce (microstructure noise). Medium windows (50–200) capture liquidity-driven mean reversion. Long windows capture session-level patterns like opening range reversals. The optimal window varies by instrument liquidity and trading frequency.
Usage Cases
- Mean reversion entry: Enter a reversal trade when reversal_probability is high and price has moved significantly in one direction. The high reversal score confirms that past moves in this instrument tend to revert, increasing the edge of the mean-reversion bet.
- Momentum filter: Avoid momentum strategies when reversal_probability is high — the tendency to revert works against trend-following. Use as a regime filter: high reversal_probability → mean-reversion strategies; low → momentum strategies.
- Market-making parameter tuning: High reversal_probability means inventory risk is lower (positions tend to mean-revert). Market makers can hold larger inventories. Low reversal_probability means positions may trend away — reduce inventory limits.
- Intraday session timing: Reversal probability varies systematically through the trading day. It's typically highest mid-day (range-bound, mean-reverting) and lowest near open/close (directional, trending). Time your strategy rotation accordingly.
YAML Definition
name: intraday_reversal_probability
description: Intraday reversal probability score - estimates likelihood that recent
directional movement will reverse
category: mean_reversion
version: v0.9.0 (Beta)
dimension: signal
status: Pre-release
required_inputs:
- returns
output_column: reversal_score
output_description: Estimated probability of near-term price reversal (0-1)
parameters:
window:
type: integer
description: Window for negative autocorrelation strength measurement
required: false
default: 50
constraints:
min: 5
max: 1000
formula: neg_autocorr(returns, window)