2026-04-10 –, Glatt (Y22-F-68/70)
Public blockchains are often described as passive settlement layers, while “real” price discovery is assumed to occur on centralized exchanges. This talk challenges that view.
Using the complete set of Ethereum transactions from 2022–2023, we study how execution behavior on-chain contributes to price discovery across decentralized and centralized markets. We introduce a simple but powerful concept—execution urgency—which measures how strongly a wallet appears to “want” its transaction executed. Urgency is inferred from observable choices such as gas fees paid, transaction timing, transaction size, routing paths, and the use of rare assets or specialized smart contracts.
We show that execution urgency is not random. Instead, it consistently captures economically meaningful behavior. A data-driven decomposition reveals two intuitive components: one related to execution intensity (how costly, large, or time-sensitive a transaction is), and another related to structural rarity (how unusual the assets, contracts, or counterparties involved are).
Most importantly, we find that high-urgency on-chain transactions systematically precede lasting price changes on centralized exchanges. These transactions explain a disproportionate share of permanent price movements, indicating that informed activity often originates on-chain before being reflected elsewhere. Using rolling-window machine-learning models, we further show that urgency-based signals predict short-horizon returns and cross-venue order-flow imbalances.
Our findings suggest that Ethereum’s execution layer is not merely infrastructure—it actively participates in information transmission. As financial activity increasingly migrates to programmable settlement systems, execution urgency emerges as a core microstructure primitive for blockchain-based markets.
Urgency is structured, not noise
Transactions cluster into two clear patterns:
(i) fast, costly, and large executions, and
(ii) rare or unconventional interactions with the blockchain.
Small share, big impact
A relatively small fraction of high-urgency transactions accounts for a large share of permanent price changes observed on centralized exchanges.
On-chain activity moves prices elsewhere
On-chain urgency predicts short-term returns and order-flow imbalances across venues—suggesting information flows from Ethereum outward, not only the other way around.
Machine learning confirms economic intuition
Models trained on urgency features outperform benchmarks in predicting short-horizon market movements, especially during volatile periods.
Arzé Karam is a researcher at Durham University Business School where she also leads a quantitative research center. She works with academic and industry collaborators on Ethereum execution, market structure, and on-chain data. She has a background in financial economics, quantitative finance and quantum finance and focuses on how trading behavior affects prices, liquidity, and execution design.
Her recent work sits at the intersection of Ethereum, market microstructure, and large-scale data analysis, with an emphasis on how real user behavior reveals information through fees, timing, routing, and smart-contract interactions. She studies how these on-chain signals propagate into prices and liquidity across decentralized and centralized markets.
Working with full-history Ethereum transaction data, Arzé and her co-authors develop practical metrics to measure execution urgency, information flow, and market stress, and uses machine-learning tools to evaluate their predictive value. Her research speaks directly to builders interested in execution design, DeFi market structure, MEV dynamics, and the role of programmable settlement layers as financial activity increasingly moves on-chain.