Skip to content

Order Book Dynamics

Difficulty advanced

Overview

The order book is the central mechanism of price discovery in modern electronic markets. Understanding its dynamics is essential for execution quality and short-term alpha.

cumulative volume midprice lower price ← → higher price bid side (buy orders) ask side (sell orders) ↕ spread
cumulative depth at each price level · taller staircase on one side = order-flow imbalance · the gap at midprice is the bid-ask spread

Order Book Structure

┌──────────────────────────────────────────┐
│  ASKS (Sell Orders) - Price Ascending     │
├────────────┬────────────┬────────────────┤
│  Price     │  Quantity  │  Orders        │
├────────────┼────────────┼────────────────┤
│  150.15    │  200       │  3             │
│  150.12    │  500       │  7             │
│  150.10    │  800       │  12            │
│  150.08    │  300       │  5             │
│  150.05    │  1000      │  15            │
│  150.03    │  400       │  8    ← Best Ask│
├────────────┴────────────┴────────────────┤
│  SPREAD: $0.01                            │
├────────────┬────────────┬────────────────┤
│  Price     │  Quantity  │  Orders        │
├────────────┼────────────┼────────────────┤
│  150.02    │  800       │  10   ← Best Bid │
│  150.00    │  1500      │  20            │
│  149.98    │  600       │  8             │
│  149.95    │  2000      │  25            │
│  149.92    │  1200      │  15            │
│  149.90    │  900       │  12            │
├────────────┴────────────┴────────────────┤
│  BIDS (Buy Orders) - Price Descending    │
└──────────────────────────────────────────┘

Order Book Events

Order Arrival

New limit order added to the book at a specific price level.

Order Execution

Market order matches against resting limit order(s).

Order Cancellation

Existing order removed from the book.

Order Modification

Order quantity or price changed.

Order Flow Imbalance

Predictive Power

OFI (Order Flow Imbalance) predicts short-term price changes:

ΔP_{t+1} = α × OFI_t + ε_t

Positive OFI → upward pressure
Negative OFI → downward pressure

where: ΔP_{t+1} next-period price change · OFI_t order-flow imbalance over the current window (Δbid-size − Δask-size, net of cancels) · α regression slope · ε_t residual. does: turns book changes into a short-horizon directional signal. Used as both a market-making skew input and a stand-alone alpha at the millisecond-to-second horizon.

Market Impact Models

Linear Impact

Impact = η × (Order Size / ADV)

Where η is a constant (~0.1-1.0)

where: Impact price displacement caused by the trade · η linear impact coefficient (~0.1-1.0) · Order Size shares to execute · ADV average daily volume. does: rough first-order cost estimate for small orders. Used in pre-trade TCA — empirically too steep for orders > 1% ADV; switch to the square-root model.

Square Root Impact (Almgren-Chriss)

Impact = σ × Y × √(Order Size / ADV)

Where Y ≈ 0.5-1.0 (market-specific)

where: Impact expected price move from execution · σ daily volatility · Y market-specific impact constant · Order Size / ADV participation rate. does: empirical regularity across asset classes — impact scales with √(participation), not linearly. Drives execution-algo schedules: split a large order across enough time so participation stays in the regime where this estimate holds.

Temporary vs. Permanent Impact

Total Impact = Temporary + Permanent

Temporary: Reverts after trade completes (spread cost, liquidity)
Permanent: Stays (information content of trade)

where: Total Impact observed price move from trading · Temporary reverts after the order finishes · Permanent price discovery from information leakage. does: distinguishes liquidity cost (temporary) from information cost (permanent). Used in TCA and impact-model calibration — execution algos can reduce temporary impact by going slower, but only stealth (icebergs, dark) reduces permanent.

Queue Position

Price-Time Priority

At each price level, orders are filled FIFO (first in, first out).

Order Book Metrics

Microprice

Microprice = (BidPrice × AskSize + AskPrice × BidSize) / (BidSize + AskSize)

Weighted average of bid and ask, weighted by size
Better predictor of next price move than midpoint

where: BidPrice, AskPrice top-of-book quotes · BidSize, AskSize top-of-book quantities. does: size-weighted fair value — leans toward the thicker side because that side is harder to consume. Used as the fair-value reference in market-making models and as a short-horizon directional signal.

Key Insights

  1. Order Book Is Dynamic — Changes in milliseconds
  2. Queue Position Matters — Being first in queue = better fills
  3. Hidden Liquidity — Not all liquidity is visible
  4. Information Content — Order flow contains information about future prices
  5. Impact Is Real — Large orders move the market
  6. Speed Is an Edge — Latency arbitrage is real (and controversial)

Next Steps