Order Book & Microstructure

How the Simulated Order Book Works

PatternFoundry's order book is a full limit order book simulation - not a simple random walk. Multiple autonomous agents submit, cancel, and modify orders in real-time, creating emergent price behavior that mirrors real market microstructure.

The order book maintains:

• Bid side - All resting buy orders, sorted by price (highest first)
• Ask side - All resting sell orders, sorted by price (lowest first)
• Trade tape - Record of all executed trades with time, price, size, and aggressor side

When an aggressive order (market order or limit order that crosses the spread) arrives, it matches against resting orders at the best available price. If the aggressive order is larger than the resting liquidity at the best price, it "walks the book" - filling at progressively worse prices until fully filled. This creates realistic slippage.

Agent Types

Four types of simulated agents interact in the order book:

Market Makers

Market makers continuously quote both sides of the book, providing liquidity. They:

• Maintain bid and ask quotes within a configurable spread of the mid-price
• Adjust quotes based on inventory (skew toward reducing position)
• Widen spreads during high volatility (uncertainty)
• Cancel and replace quotes as the mid-price moves

Market makers are the primary source of resting liquidity and are responsible for the bid-ask spread.

Institutional Agents

Institutional agents represent large directional traders (hedge funds, asset managers). They:

• Submit large orders that move price
• Trade with persistence - once they start buying/selling, they continue in the same direction (modeling TWAP/VWAP execution)
• Create sustained directional pressure that drives trends
• May use limit orders to minimize market impact (iceberg behavior)

Retail Agents

Retail agents represent individual traders. They:

• Submit small, randomly-directed orders
• Add noise and background liquidity
• Have no persistence - each order is independent
• Primarily use market orders (aggressive)

Momentum Agents

Momentum agents chase price movement. They:

• Buy when price is rising, sell when price is falling
• Create positive feedback loops that accelerate moves
• Contribute to trend continuation and stop cascades
• Activate more strongly during high-volatility periods

Hawkes Self-Excitation (Trade Clustering)

Real markets exhibit trade clustering - trades tend to arrive in bursts rather than uniformly. PatternFoundry models this using a Hawkes process, a self-exciting point process where each event increases the probability of subsequent events.

The intensity (rate of order arrivals) at time t is:

λ(t) = λ₀ + Σ α · e^(-β(t - tᵢ))

Where:

• λ₀ (Base Intensity) - Background arrival rate when no excitation is present
• α (Alpha) - Jump in intensity caused by each trade
• β (Beta) - Exponential decay rate of excitation
• tᵢ - Times of previous trades

This creates realistic bursts of activity: a single trade triggers more trades, which trigger even more, creating cascading activity that naturally decays back to baseline. This is visible in the tape as clusters of rapid trades followed by quiet periods.

Price Impact from Order Flow

Price impact in PatternFoundry follows realistic market microstructure principles:

• Temporary impact - Large aggressive orders push price away from equilibrium. The impact is proportional to order size relative to available liquidity.
• Permanent impact - Sustained directional flow (institutional persistence) shifts the equilibrium price permanently.
• Book depletion - When one side of the book is consumed faster than it's replenished, price moves rapidly in that direction (vacuum/air pocket).
• Spread dynamics - After a large trade depletes liquidity, the spread widens temporarily until market makers replenish quotes.

Liquidity Pools (Support/Resistance)

Liquidity pools model the tendency for orders to accumulate at specific price levels, creating support and resistance:

• Pools are placed at psychologically significant levels (round numbers, previous highs/lows)
• They attract price via the Strength parameter - higher strength means price is more likely to gravitate toward pool levels
• When price reaches a pool, the concentrated liquidity can either absorb the move (support/resistance holds) or, if overwhelmed, create a breakout with acceleration (stop cascade)
• The Count parameter determines how many pools exist in the visible range

This creates realistic chart patterns: price bouncing between levels, false breakouts, and genuine breakouts with follow-through.

DOM Panel Interpretation

Reading the DOM (Depth of Market) panel effectively: