White Paper

Vela Exchange — Monolith Research Vol. 2, by Arya Somu.

The Vela Exchange white paper is the authoritative reference for the system’s architecture, mathematical foundations, and performance methodology. It is published through Monolith Research, the research division of Monolith Systematic LLC.

Read the White Paper on SSRN

Vela Exchange — Monolith Research Vol. 2. SSRN Abstract ID: 6579199.

Paper Details

Field	Value
Title	Vela Exchange
Series	Monolith Research, Vol. 2
SSRN Abstract ID	6579199
Author	Arya Somu
Affiliation	Monolith Systematic LLC
Published	April 2026

Abstract

Vela Exchange is a high-performance verifiable central limit order book (CLOB) spot exchange. The paper presents the complete design: a single-threaded Rust matching engine with microsecond latency, a novel market maker credit system for capital-efficient multi-market quoting, an optimistic-ZK verification layer for trustless state transition proofs, and a forced inclusion mechanism for anti-censorship guarantees.

Benchmarks show 1.38 μs median match latency and 725,000 operations per second, representing a 5.8× throughput improvement over the Pulse baseline with a 73% reduction in p99.9 tail latency.

Sections

1. Motivation and CEX vs. DEX Tradeoffs

The paper opens with a structured analysis of the centralized vs. decentralized exchange landscape. It quantifies the latency and throughput gap between CEXes and on-chain DEXes, and identifies the specific trust assumptions each model requires from users.

The key insight: the tradeoff is not fundamental. With off-chain matching, cryptographic authentication, and a verifiable state transition layer, it is possible to approach CEX performance while providing meaningful trustlessness guarantees.

2. Architecture Deep Dive

Full description of the six-crate Rust workspace: types, engine, state, api, committer, and zkvm. Includes data flow diagrams, interface definitions, and rationale for key design decisions.

Notable topics: why a single-threaded engine outperforms naive parallelism for a CLOB, the CoW cache execution model, and the relationship between the MPT state layer and the zkvm prover.

3. Market Maker Credit System

The credit system allows market makers to quote across multiple markets with a single collateral deposit. The paper derives the credit utilization formula:

total_quoted_value = Σ(price_i × size_i)
credit_utilization = total_quoted_value / (deposited × credit_ratio)

It proves that atomic auto-cancel — firing at the moment a fill would push utilization past 1.0 — is necessary for the invariant to hold, and describes the implementation that achieves this within the single-threaded engine.

4. Optimistic-ZK Design

The paper describes the optimistic execution model: batches are assumed correct by default, and any party can challenge by running the zkvm prover against the published DA batch. A fraud proof is generated if the prover’s state root diverges from the published root.

This section covers the verify_execution() algorithm, the ZkvmInput and FraudProof data structures, and the planned on-chain verifier contract design.

5. Benchmarking Methodology

Benchmarks are run using Criterion.rs with statistical rigor: 100 sample runs, outlier detection, and comparison against a baseline. The paper describes the benchmark harness, the workload generator (uniform random order stream across all price levels), and the hardware profile.

All benchmarks are reproducible from the open-source repository.

6. Comparison to Pulse, dYdX, Vertex, and Serum

Structured comparison of Vela against four reference systems:

• Pulse — the baseline CLOB implementation used for benchmarking
• dYdX v4 — off-chain matching with on-chain settlement, Cosmos-based
• Vertex Protocol — integrated AMM + CLOB on Arbitrum
• Serum / OpenBook — on-chain CLOB on Solana

Comparison axes: match latency, throughput, trust model, DA approach, on-chain settlement, fraud proof support.

Citation

If you reference the Vela Exchange white paper in academic or technical work, please use:

Somu, Arya. "Vela Exchange." Monolith Research, Vol. 2. Monolith Systematic LLC, April 2026.
Available at SSRN: https://ssrn.com/abstract=6579199

Related Resources

Architecture Overview

Codebase-level walkthrough of the six crates.

Matching Engine

Engine internals: price-time priority, CoW cache, credit integration.

zkVM and Fraud Proofs

Optimistic-ZK design and verify_execution() algorithm.

MM Credit System

Capital-efficient multi-market quoting math and implementation.