Research
Chain Reactions: How Nonce Collisions in ECDSA Compromise Polygon MEV Searchers
ECDSA signatures are only secure if each signature uses a fresh, unpredictable nonce k. The authors argue that Polygon MEV searchers, forced to respond inside sealed-bid auction windows, reuse nonces…

- arXiv
- 2605.21498
- Published
- 2026-05-03
- Authors
- Yash Madhwal, Andrey Seoev, Raffaele Della Pietra, Anastasiia Smirnova, Yury Yanovich
AI summary
Overview
Research area: Applied cryptography and blockchain security — specifically ECDSA nonce generation in the Polygon MEV (maximal extractable value) ecosystem.
Technical level: Intermediate. The paper is short and builds its argument from basic ECDSA signing algebra, so readers with some familiarity with elliptic-curve signatures will follow it easily; the case-study and mitigation sections are accessible to a general blockchain audience.
Scope: The paper documents repeated and cross-wallet ECDSA nonce reuse among Polygon MEV searchers and shows, via a compact linear-system formulation and on-chain evidence, that such patterns permit complete private key recovery by a passive observer.
What This Paper Is About
ECDSA signatures are only secure if each signature uses a fresh, unpredictable nonce k. The authors argue that Polygon MEV searchers, forced to respond inside sealed-bid auction windows, reuse nonces as a latency optimization — and that this turns public blockchain data into a key-recovery pipeline. The goal is to formalize the resulting attacks (including the more dangerous cross-wallet case) and to present concrete on-chain evidence that these patterns occur in production.
Key Contributions
- A compact linear-system formulation that unifies three compromise conditions for ECDSA: single-wallet nonce reuse (C1), linear nonce relations of the form
k_i = α_i·k_0 + β_i(C2), and cross-wallet nonce collisions (C3). - A matrix formulation (Equation 3) showing that two cross-wallet nonce collisions between two private keys
d_Aandd_Byield exactly enough independent linear equations to solve for both keys simultaneously modn. - Concrete on-chain evidence of nonce reuse on Polygon, reporting four addresses grouped into two "same
k" collisions, plus a description of three observed failure modes in transaction logs: constantkvalues, sequential reuse, and cross-wallet collisions. - A mitigation roadmap spanning immediate fixes (RFC 6979 deterministic nonce generation, monitoring for repeated
rvalues, key rotation, historical audits) and longer-term measures (signing-component audits, developer education, and consideration of Schnorr or EdDSA).
Main Findings
- Nonce reuse is structurally exploitable: Because
s ≡ k⁻¹(e + x·r) (mod n), the signature depends linearly on the private keyxoncekis fixed, so any structure inkbecomes "a direct pipeline tox." - Single reuse is total compromise: Two signatures sharing a nonce produce matching
rvalues, and subtracting the signing equations yieldsk ≡ (e₁ − e₂)(s₁ − s₂)⁻¹, thenx ≡ (s₁k − e₁)r⁻¹— "one reuse, complete compromise." - Cross-wallet collisions are the critical case (C3): If two different private keys sign with the same nonce
k₁, eliminatingk₁gives one linear relation betweend_Aandd_B; a second collision with noncek₂gives a second independent equation, enough to solve for both keys. - Cascading rather than isolated risk: Once one wallet is compromised, every wallet that collided with it is exposed, so the pattern is described as creating ecosystem-wide risk chains, modeled as a bipartite graph of keys and nonces.
- The driver is protocol-induced latency pressure: The shift from spam-based Priority Gas Auctions to sealed-bid FastLane auctions (Polygon Atlas) requires bids within 250 ms windows, which the authors say pushes searchers to reuse
kas a latency optimization. - Evidence is observational and historical: Using custom Python tools on "a representative sample of Polygon blocks," the authors report constant
kvalues, sequential reuse, and cross-wallet collisions in transaction logs, noting searchers abandon the practice after exploitation but leave permanent on-chain trails. The sample size (number of blocks scanned) is not reported. - Two concrete collision pairs are listed: Addresses
0xc1e9…73fdwith0x765a…2200, and0x722f…2666with0xd2ad…448f. - Full recovery was demonstrated: "Our Python implementation confirms complete key recovery," though the paper does not report per-address recovered balances or losses.
- Existing knowledge, persistent failure: The authors stress the attack is a well-understood implementation error, not a new cryptographic break.
Methodology in Plain English
The authors start from the standard ECDSA signing equation and treat each observed signature as one linear equation in the unknown private key and the unknown nonce, all modulo the curve order n. They then ask what happens when the nonces are not independent: if two signatures share a nonce, or if nonces follow a known linear pattern, the equations can be combined to cancel unknowns. For the cross-wallet case, they write the equations for two keys signing with the same nonce, cancel the nonce, and show that two such collisions give a 2×2 linear system solvable by ordinary algebra modulo n. Separately, they scan a sample of Polygon blocks with custom Python tooling to look for repeated r values and matching nonces across addresses, then verify key recovery with an implementation.
Why This Matters
Impact on research: The paper reframes nonce misuse as a systemic, multi-account phenomenon rather than a per-wallet bug. By treating key–nonce collisions as a graph, it gives security researchers a tractable structure for reasoning about cascading compromise across an ecosystem, and it links protocol design (auction latency) to cryptographic failure — a connection that pure cryptanalysis papers rarely make.
Real-world applications:
- Wallet and signing-library audits: checking whether signing services ever emit repeated
rvalues, within or across accounts. - MEV infrastructure hardening: searchers and relay/builder operators can treat nonce generation as a security-critical path rather than a performance detail.
- Chain monitoring and forensics: block explorers, analytics firms, and exchanges could flag repeated
rvalues across wallets as a compromise indicator. - Incident response: after a suspected key leak, cross-wallet collision history tells a victim which counterparties are also exposed.
Industry relevance: The finding directly implicates latency-sensitive trading infrastructure on Polygon, but the same pressure exists on any chain with competitive ordering. The paper's central warning — that "a single implementation error compromises multiple accounts simultaneously" — applies to any production system where speed is prioritized over entropy quality and where a single bad signing path is shared across many keys.
Future Directions
- Full-chain prevalence measurement: The paper scans only a representative sample; a systematic, chain-wide census of repeated
rvalues on Polygon (and comparison with other chains) would establish how widespread the pattern is. No prevalence or frequency figures are reported. - Automated detection tooling: Turning the linear-system formulation into continuous monitoring that alerts on cross-wallet nonce collisions in real time, rather than post-hoc analysis.
- Migration evaluation: Assessing the practical costs and benefits of moving signing infrastructure to Schnorr or EdDSA, and of enforcing RFC 6979 across searcher tooling.
- Graph-level risk modeling: Developing the bipartite key–nonce graph into a metric for ecosystem exposure, so that a single compromise can be mapped to all downstream wallets at risk.
- Release of artifacts: The authors state they will publish their analysis and key recovery scripts upon acceptance, which would enable independent replication.
Target Audience
Blockchain security researchers and cryptographers studying real-world signature failures; MEV searchers, builders, and relay operators whose signing infrastructure is directly implicated; wallet and custody providers responsible for nonce generation; smart-contract and protocol auditors; and incident responders and forensic analysts who need to recognize cross-wallet collision patterns in on-chain data. Developers new to ECDSA will also find the compact derivation of the reuse attack a useful, self-contained worked example.
Authors’ abstract
ECDSA signatures form the bedrock of blockchain transaction authentication, yet their security critically depends on proper nonce generation. We uncover a critical vulnerability in the Polygon MEV ecosystem: systematic nonce reuse that enables complete private key recovery. Analyzing on-chain data reveals that searchers, driven by the need for sub-second response times in sealed-bid auctions, employ predictable nonce patterns. These patterns create linear relationships…
The full abstract on arXiv