A high-frequency trading (HFT) style Limit Order Book implemented in Java, designed to demonstrate lock-free concurrency, zero-garbage collection (Zero-GC) in the hot path, and low-latency architecture.

• Object Pooling: Pre-allocated Order objects (100k+) to avoid new keyword allocations during trading.
• Primitive Collections: Uses fastutil (e.g., Long2ObjectOpenHashMap) to avoid boxing/unboxing overhead for Order IDs.
• Intrusive Linked Lists: Custom OrderLevel doubly-linked list uses Order objects themselves as nodes, eliminating Node wrapper allocations.

• Lock-Free Reads: Uses StampedLock (Optimistic Read) for L1 Market Data (Best Bid/Ask) access.
• Fine-Grained Locking: Matches at specific price levels are protected by ReentrantLock per level, allowing concurrent matching at different prices.
• Atomic Operations: Uses ConcurrentSkipListMap with atomic compute() operations to handle order book updates without global locks.
• Deadlock Prevention: Strict locking hierarchy and "Mark-as-Removed" patterns to safely handle concurrent cancellations and matching.

• Data Structures: ConcurrentSkipListMap for sorted price levels (O(log n)) and Long2ObjectOpenHashMap for O(1) order lookups.
• Throughput: Designed to sustain 100k+ orders/second on standard hardware.
• Latency: Sub-millisecond P99 latency for order matching.

• OrderPool: A thread-safe, array-based stack for recycling Order objects.
• OrderBook: The central data structure managing Bids (Descending) and Asks (Ascending).
• OrderLevel: A custom linked list representing a queue of orders at a specific price.
• MatchingEngine: Implements the Price-Time Priority (FIFO) matching algorithm.
• OrderBookController: Exposes the system via a REST API (Javalin).

• Language: Java 17+
• Build Tool: Gradle
• Web Server: Javalin (Lightweight REST API)
• UI: Streamlit (Python) for real-time visualization
• Collections: fastutil, Java Concurrent Utils

• Java 17+
• Python 3.8+ (for UI)

./gradlew build

Starts the CLOB engine and REST API on port 8080.

./gradlew runApiServer

Starts the Streamlit dashboard on port 8501.

pip install -r ui/requirements.txt
streamlit run ui/streamlit_app.py

The system exposes a REST API at http://localhost:8080.

POST /api/orders
{
  "side": "BUY",
  "price": 10500,
  "quantity": 100
}

Note: Price is fixed-point (10500 = $105.00)

To run the JMH microbenchmarks (if configured):

./gradlew jmh

src/main/java/com/hft/clob/
├── api/            # REST API Controller
├── core/           # Core Data Structures (Order, Book, Pool)
├── engine/         # Matching Logic
└── ApiServer.java  # Application Entry Point
ui/                 # Streamlit Dashboard