Queue Position Imbalance

Quick Reference

Property	Value
Dimension	signal
Category	order_flow
Version	v0.9.0 (Beta)
Output Column	queue_position_imbalance

Queue position adjusted order book imbalance with position weighting

Formula

CDM Inputs

Column	CDM Table	Description
best_bid_size	cdm_*	CDM source table
best_ask_size	cdm_*	CDM source table

Parameters

Parameter	Type	Default	Description
position_weights	list	[10, 9, 8, 7, 6, 5, 4, 3, 2, 1]	Position weights (closer to mid price = higher weight)
normalization_window	integer [10, 1000]	100	Window for z-score normalization

Output

Column: queue_position_imbalance

Normalized queue position weighted order book imbalance

Market Intuition & Trading Rationale

Queue position imbalance computes a position-weighted order book imbalance, normalized via rolling z-score. Unlike simple depth_imbalance (which weights all shares equally), queue position weights each share by its proximity to the front of the queue — the first share in line gets the highest weight, the last gets the lowest. This captures a critical microstructure reality: the marginal share at the front of the queue is more informative than the marginal share at the back.

The multi-step computation extracts sizes at each level, computes per-level imbalance ratios, applies position-based weights, and normalizes via rolling z-score. The default position weights (10 down to 1) heavily favor the touch. This makes queue_position_imbalance more responsive than depth_imbalance — it reacts faster to changes at the front of the queue while being less influenced by deep-book positioning.

The rolling z-score normalization makes this feature self-calibrating. A value of +2 means the current queue imbalance is two standard deviations above its recent mean — statistically significant buying pressure at the front of the queue.

Usage Cases

• Micro-price prediction: queue_position_imbalance predicts short-term mid-price moves more accurately than raw depth_imbalance because it focuses on the most informative part of the book (the front of the queue). Trade in the direction of significant (|z| > 1) imbalances.
• HFT signal: The position weighting makes this feature responsive enough for HFT — changes at the touch propagate immediately into the signal. Use for sub-second directional predictions.
• Adverse selection detection: When queue_position_imbalance is strongly positive but price is falling, informed sellers may be absorbing the bid-side liquidity. The imbalance shows buying support, but it's being consumed — a bearish divergence.

YAML Definition

name: queue_position_imbalance
description: Queue position adjusted order book imbalance with position weighting
category: order_flow
version: v0.9.0 (Beta)
dimension: signal
status: Pre-release
required_inputs:
  - best_bid_size
  - best_ask_size
output_column: queue_position_imbalance
output_description: Normalized queue position weighted order book imbalance
tags:
  - orderbook
  - lob
  - imbalance
  - microstructure
  - array
  - advanced
parameters:
  position_weights:
    type: list
    description: Position weights (closer to mid price = higher weight)
    required: false
    default:
      - 10
      - 9
      - 8
      - 7
      - 6
      - 5
      - 4
      - 3
      - 2
      - 1
  normalization_window:
    type: integer
    description: Window for z-score normalization
    required: false
    default: 100
    constraints:
      min: 10
      max: 1000
steps:
  - primitive: TRANSFORM
  op: array_extract
  params:
    array: best_bid_size
    field: size
    output_col: best_bid_size
  - primitive: TRANSFORM
  op: array_extract
  params:
    array: best_ask_size
    field: size
    output_col: best_ask_size
  - primitive: TRANSFORM
  op: array_elementwise_divide
  params:
    numerator:
      op: sub
      a: bid_sizes
      b: ask_sizes
    denominator:
      op: add
      a: bid_sizes
      b: ask_sizes
  output: imbalance_per_level
  - primitive: TRANSFORM
  op: array_wavg
  params:
    array: imbalance_per_level
    weights: $position_weights
  output: weighted_imbalance
  - primitive: WINDOW
  op: rolling_zscore
  params:
    window: $normalization_window
    a: weighted_imbalance
  output: queue_position_imbalance