Ultra-Fast Market Data Processing

Parse NASDAQ ITCH data at institutional speeds. Reduce backtest times from days to hours. Built in Rust for institutional trading infrastructure.

Request Demo View Benchmarks

213.7 M

msg/s framing

93.9 M

msg/s decode

499.7 M

msg/s 8-vCPU

Validated checksum

Your Trading Research Is Bottlenecked by Slow Data Processing

Historical Feeds Are Large

Historical ITCH sessions contain hundreds of millions of binary messages and require efficient framing, decoding, and validation.

Parsing Adds System Cost

Allocation-heavy parsing can add memory and CPU work at high message rates. Direct field access keeps the hot path focused.

Research Needs Complete Data

Message-level access helps research workflows analyze order, cancel, execution, and trade events without sampling the feed.

Memory-Mapped Parsing for Historical NASDAQ Data

Lunary parses NASDAQ TotalView-ITCH 5.0 feeds directly from memory-mapped historical data and keeps per-message overhead low.

Median results across 10 full-dataset runs
Validated against an independent reference decoder
Message boundaries are kept aligned to avoid partial reads
Benchmarks exclude disk and network overhead

Built in Rust with memory-mapped parsing, direct binary decoding, and deterministic low-overhead execution.

Verified NASDAQ benchmark results

Mode	Throughput	Data Rate
Single Thread, Framing	213.7 M msg/s	6.6 GB/s
8 vCPUs, Full Decode	499.7 M msg/s	15.4 GB/s

See full benchmark details

Built for Quantitative Trading Operations

Strategy Backtesting

Process years of historical order book data in hours, not days. Test more strategies, iterate faster, identify alpha before competitors.

Verified benchmark: 93.9 M msg/s single-thread full decode on the tested historical session.

Real-Time Order Book Reconstruction

Reconstruct full depth-of-book state with nanosecond-level latency for live trading strategies that depend on order flow.

Measured processing cost: 5.2 ns/message for framing and 9.9 ns/message for full decode.

Market Microstructure Research

Analyze complete order flow without sampling artifacts. Full message-level access for academic research and proprietary analysis.

Validation scope: 293,989,079 messages checked against an independent reference decoder.

Market Data Infrastructure

Normalize and enrich ITCH feeds for downstream analytics, compliance, and reporting systems with enterprise-grade reliability.

Supported workflow: memory-mapped historical input with direct binary decoding and structured export options.

Architecture Designed for Speed

Zero-Copy Parsing

Operates directly on ingested buffers, eliminating memory allocation overhead that compounds at high message rates.

SIMD Vectorization

AVX2 instructions process multiple data elements simultaneously, maximizing CPU pipeline utilization.

Lock-Free Concurrency

Wait-free data structures enable parallel processing without synchronization bottlenecks.

Cache Optimization

Memory layouts aligned to cache boundaries with predictable access patterns minimize latency.

Technical Documentation

Verified Performance

All performance claims are reproducible using our open-source benchmark suite.

Testing methodology includes:

Official NASDAQ historical ITCH files (production data, not synthetic)
Controlled hardware environments with documented specifications
Statistical analysis of latency distributions across all message types
Long-duration stability validation (72+ hour continuous operation)

Our open-source version is available on GitHub for evaluation.

View Benchmark Methodology GitHub Repository

Start Evaluating Today

Evaluate Our Approach

Download the open-source version and verify performance claims with your own data.

GitHub Repository

Production Deployment

Schedule a technical consultation to discuss your infrastructure requirements and evaluation timeline.

Request Demo