Kremena Valkanova

I am a microeconomic theorist studying how people, algorithms, and societies make decisions, and how the choice environment shapes outcomes. My research examines what decisions reveal about preferences, attention, and cognition, and how decision-makers learn and adapt, particularly when choices are stochastic or seemingly inconsistent. I also study how to design institutions that promote fair, robust, and timely collective decision-making.

I am currently a Research Scientist at Parity Technologies, where I work on voting system design for Polkadot governance and on blockchain economic security. I completed my Ph.D. in Economics at the University of Zurich in 2021 under the supervision of Nick Netzer and Jakub Steiner, followed by a postdoctoral position on mathematical democracy at ETH Zurich with Hans Gersbach.

Resume

Research

Working Papers

Optimal Dynamic Majority-Quorum Rules: Theory and Evidence

with Bhargav Nagaraja Bhatt and Jonas Gehrlein, September 2026

We derive an optimal stopping rule for real-time vote aggregation: it determines when to stop collecting votes and make a majority decision before a deadline. We also provide a tractable approximation and evaluate the rule on Polkadot governance data.

Markov Stochastic Choice

September 2024 (First version: November 2020)

We propose a Markovian model of sequential comparison to study how item arrangement affects choice, identifying when rearrangements are irrelevant and when observed choices reveal underlying consideration sets and decision processes.

Voting with Random Proposers: Two Rounds May Suffice

with Hans Gersbach, First Version: October 2024, This Version: August 2026, R&R in Games and Economic Behavior

We study a randomized agenda-setting procedure that curbs manipulation and quickly selects the Condorcet winner, with applications to committees, legislatures, and decentralized governance.

Publications

Revealed Preference Domains from Random Choice

2024, Games and Economic Behavior, 147, pp. 288-304.

We define a new property of ordinal random utility models, exclusiveness, and show that many classical and novel preference domains satisfy it, enabling their direct identification from stochastic choice data.

Irrelevant Alternatives Bias Large Language Model Hiring Decisions

with Pencho Yordanov (2024), In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 6899–6912, Miami, Florida, USA. Association for Computational Linguistics.

We document a robust attraction effect in GPT-3.5 and GPT-4 hiring decisions, demonstrating that LLMs reproduce a classic human choice bias.

Fictitious Play in Networks

with Christian Ewerhart (2020). Games and Economic Behavior, 123, pp. 182-206.

We characterize when fictitious play converges in networks of bilateral games, proving fast convergence in zero-sum networks and identifying structural conditions for broader classes.