# Signals and Spoils: Speculative Oracle Extractable Value in the Era of Cross-Chain Interoperability

Crypto.world

> Aave and similar lending protocols become liquidatable only when an oracle price update changes a borrower's health factor, so whoever acts immediately after an oracle update captures the liquidation…

**arXiv**: 2606.03434

**Published**: 2026-06-02

**Authors**: Hasret Ozan Sevim, Christof Ferreira Torres

## AI summary

## Overview

**Research area:** Blockchain security and DeFi market microstructure — specifically Maximal Extractable Value (MEV) and its oracle-driven subset, Oracle Extractable Value (OEV), studied across Ethereum and the Layer-2 chains Arbitrum, Base, and Optimism.

**Technical level:** Intermediate to Advanced. The paper assumes familiarity with Chainlink's Off-Chain Reporting (OCR) architecture, Aave V3 health-factor liquidations, and rollup block production, though its core argument can be followed by a well-informed non-specialist.

**Scope (one sentence):** An empirical measurement study of whether liquidators on Layer-2 chains speculatively pre-position liquidation transactions ahead of Chainlink price updates, and whether oracle update timing differences across chains create predictable cross-chain exploitation windows.

## What This Paper Is About

Aave and similar lending protocols become liquidatable only when an oracle price update changes a borrower's health factor, so whoever acts immediately after an oracle update captures the liquidation profit. Because Arbitrum, Base, and Optimism have no public mempool, searchers cannot see pending oracle updates and instead spam near-identical liquidation transactions, hoping one lands in the right block. The paper measures how common this "speculative" behavior already is in production, and tests whether an oracle update visible on one chain statistically predicts a coming update — and thus a liquidation opportunity — on another chain.

## Key Contributions

1. **First empirical study of speculative OEV "in the wild."** The authors build a detection methodology based on transaction placement relative to oracle price updates and apply it to real Aave liquidations, showing speculative liquidators already operate in production.
2. **Quantification of the cost side of speculative OEV.** The paper shows speculative liquidation strategies generate profits but also transaction spam and reverted transaction activity, framing this as a scalability and efficiency problem for Layer-2 systems.
3. **Characterization of the latency chain.** The authors link off-chain market movements (Binance second-level spot prices), DON price observations, on-chain oracle transmissions, and liquidation execution across multiple blockchains.
4. **Formalization of the cross-chain oracle exploitation window.** The paper demonstrates that Chainlink updates on Optimism can predict subsequent updates on Arbitrum and Base, making speculative cross-chain OEV extraction plausible.

## Main Findings

- **Speculative liquidators are already a majority of actors but a minority of successful transactions.** On October 10, 2025, the authors identify 64 speculative liquidators on Aave, equal to 57% of all detected liquidators, and 831 successful speculative liquidations, equal to 39% of all successful liquidations across the three chains (Arbitrum, Base, Optimism).
- **Speculative behavior appears on every chain, most heavily on Base.** Table 7 reports 12 speculative liquidators on Arbitrum (42.86% of that chain's liquidators), 34 on Base (64.15%), and 18 on Optimism (56.25%). By liquidation events, speculative activity accounts for 447 on Arbitrum (50.45%), 193 on Base (42.60%), and 191 on Optimism (23.38%).
- **Non-speculative liquidators win more transactions.** Only 33% of successful liquidation transactions were classified as speculative and only 39% of liquidation events, with Optimism the extreme case at 23% of transactions speculative versus 77% non-speculative.
- **Speculative actors bundle multiple liquidations per transaction.** The number of speculative liquidation events exceeds the number of speculative liquidation transactions. On Arbitrum, 60% of successful liquidation transactions are non-speculative, yet 51% of successful liquidation events are speculative.
- **Same-block liquidations occur despite no public mempool.** Base shows the highest proportion of same-block liquidations at 37.31% (169 of 453), Arbitrum 19.08% (169 of 886), and Optimism 9.79% (80 of 817). On Arbitrum, 53.73% of these are back-to-back with the oracle update (transaction distance of one), consistent with a median distance of 1.00.
- **Block production shapes transaction distance.** Base's median transaction distance is 29.50 (mean 109.22, max 1,697) and Optimism's is 37.00 (mean 55.40, max 306), while Arbitrum's is 1.00 (mean 2.52, max 14) — attributed to Arbitrum's ~250 ms block time versus roughly 2 seconds on Base and Optimism.
- **The top Arbitrum liquidator is speculative.** Address 0xb9516c655831917704A5000bA6f4EB64E816DC2d executed 215 successful liquidations (24.27%), 110 of them (12.42%) in the same block as the oracle update, with a median oracle distance of 1.00. On Base and Optimism the top liquidator by count was not speculative, but the second most successful on each chain was.
- **DON peers often submit identical observations.** The lowest share of identical observations is 15% for LUSD/USD on Arbitrum; the highest is approximately 90% for wstETH/ETH on Base, suggesting overlapping or insufficiently diverse off-chain data sources.
- **Oracle configurations vary sharply across chains.** Heartbeats in the collected configuration tables range from 3,600 to 86,400 seconds (with 1,200-second heartbeats on several Base and Optimism feeds), and deviation thresholds from 0.01% to 2.00%. The ARB/USD feed on Arbitrum produced the most observations of any feed, at 14,647.
- **SVR was Ethereum-only during the study.** Chainlink's Smart Value Recapture was operationally active only on Ethereum and had not been deployed across Layer-2 networks during the observation period.
- **Cross-chain prediction is feasible.** Independent DONs consume largely identical off-chain data nearly simultaneously but publish at different times, producing statistically predictable windows; updates on Optimism are shown to predict later updates on Arbitrum and Base.

## Methodology in Plain English

The authors pick a single, extreme day — October 10, 2025, described as the largest single-day liquidation volume in crypto-asset market history measured in USD — and reconstruct everything that happened around it.

First they collect every Aave `LiquidationCall` event across Arbitrum, Base, Ethereum, and Optimism. For each liquidation they walk the contract chain: from the Aave pool to the address provider, to the Aave oracle, to `getSourceOfAsset()` for the collateral and debt assets, and finally to the Chainlink aggregator via `aggregator()`. That yields the exact price feeds that determined each liquidation's health factor.

Then they pull all Chainlink oracle events from those feeds using archive nodes and a custom Python script built on `web3` — the `NewTransmission` events, which include the raw individual price observations submitted by each DON peer, plus transmitter identities, deviation thresholds, heartbeats, SVR compatibility, and peer lists from on-chain contracts and Chainlink APIs. The dataset comprises 63 Chainlink feeds, 12,009 price updates, and over 100,000 oracle observations linked to 2,986 Aave liquidations. They add Binance second-level spot data for ARB, OP, ETH, LINK, BTC, AAVE, and CRV as the off-chain reference.

To find speculative behavior, they take each of the 2,986 liquidation transactions and scan the preceding 10 and subsequent 10 blocks for other transactions with the same destination address and identical input calldata. A liquidation counts as speculative if such a duplicate appears in the same block or in an immediately adjacent block; a liquidator is labeled speculative if any of its successful liquidations is speculative, in which case all of its successful liquidations are labeled speculative too. Separately, they measure "same-block" liquidations by grouping oracle updates and liquidations by block number, discarding oracle updates positioned after the liquidation by transaction index, and matching oracle contract addresses.

## Why This Matters

The paper reframes oracle latency not just as a price-accuracy issue but as a cross-chain information leak: independent oracle networks that read the same exchanges but publish on different schedules hand predictive signals from one chain to another.

**Impact on research:** It opens a concrete research direction that prior work (cited as [3]) had only flagged as underexplored — cross-chain latency arbitrage in OEV markets — and provides a reusable detection methodology plus a public-day dataset linking liquidations, oracle transmissions, and raw DON observations.

**Real-world applications:**

- **Oracle design:** Informing deviation-threshold and heartbeat configurations that reduce cross-chain predictability.
- **Protocol revenue and auctions:** Extending mechanisms like Chainlink SVR, currently Ethereum-only, to Layer-2 markets so liquidation MEV is recaptured rather than leaked to searchers.
- **Liquidator and searcher strategy:** Quantifying how much of L2 liquidation flow is already speculative, and where same-block placement succeeds.
- **Layer-2 capacity planning:** Framing speculative transaction spam and reverted transactions as a blockspace-consumption problem with fee implications for ordinary users.

**Industry relevance:** The findings touch oracle providers, lending protocols, rollup operators, and MEV infrastructure builders, all of whom make design decisions about update timing, auction design, and block production that this paper shows are economically consequential.

## Future Directions

- **Does SVR on Layer-2s change the picture?** SVR was active only on Ethereum during the study; measuring whether routing L2 updates through sealed-bid auctions removes speculative same-block competition is the obvious next test.
- **How generalizable is a single day?** The analysis is anchored to October 10, 2025, an extreme stress day; whether speculative shares (64 speculative liquidators, 831 speculative liquidations) hold during calmer periods is untested.
- **Can cross-chain prediction be turned into measurable profit?** The paper shows Optimism updates statistically predict Arbitrum and Base updates, but the economic magnitude of actually executing on those windows, net of fees and failed transactions, is left open.
- **What are the mitigation trade-offs?** Lower deviation thresholds and shorter heartbeats improve price accuracy but generate more updates, more opportunities, and more spam; the paper raises the question of how to tune oracle parameters without amplifying speculative extraction.

## Target Audience

Blockchain security researchers and MEV/OEV specialists will find the core contribution; oracle engineers and protocol designers at lending platforms will care about the configuration and SVR findings; Layer-2 operators and infrastructure teams will value the blockspace and spam analysis. DeFi liquidators and quantitative trading firms will find the transaction-distance statistics directly operationally relevant, and graduate students entering MEV research will find the methodology section a workable template. Readers need comfort with Aave health-factor mechanics and oracle update semantics to get the most out of it.

## Authors’ abstract

A new form of Maximal Extractable Value (MEV), termed speculative MEV, has emerged across Layer-2 blockchains. Unlike Ethereum mainnet, many Layer-2 systems lack a public mempool, forcing extraction strategies to become probabilistic: searchers emit multiple identical transactions hoping to capture an opportunity first. This generates substantial transaction spam, increasing fees and wasting block space. We investigate speculative Oracle Extractable Value (OEV), a form of…

[The full abstract on arXiv](https://arxiv.org/abs/2606.03434)

[Read the original paper](https://arxiv.org/abs/2606.03434)

---

Canonical: https://crypto.world/paper/signals-and-spoils-speculative-oracle-extractable-value-in-the-era-of-2606.03434
