Online Learning and Economics

The workshop focuses on the intersection of online learning and economics, exploring how learning algorithms are increasingly used for decision-making in strategic economic settings such as markets and platforms. The event brings together researchers from diverse fields to discuss current challenges, share recent advances, and foster collaboration. The scope of the workshop extends beyond online learning to encompass other relevant domains of machine learning, including, for example, learning theory and reinforcement learning.

Website to the First Edition (link)

• 09:00 – 10:30 First Session of Invited Talks - Room 6

• 10:30 – 11:00 Coffee Break

• 11:00 – 12:30 Second Session of Invited Talks - Room 6

• 12:30 - 14:00 Lunch / Poster session

For details on which posters are assigned to each session, please refer to the Accepted Posters section.

Chara Podimata (MIT)

TBA

Gabriele Farina (MIT)

TBA

Nicolò Cesa-Bianchi (Università degli Studi di Milano)

Trading under constraints: Periodic strategies for blocking bandits

Abstract

In automated financial markets, trading algorithms face structural bottlenecks: quantitative funds encounter capital lockups during venue settlement cycles, while market makers navigate inventory risk via mandatory cool-down windows. Both scenarios constrain continuous execution by temporarily blocking an asset or venue from being immediately re-selected. We formalize these operational limits using the adversarial blocking bandit framework, where playing an arm renders it unavailable for a fixed number number of future rounds.

We first show that computing the optimal unconstrained dynamic policy in an adversarial market environments is NP-hard. To establish a tractable alternative, we turn to d-periodic policies, which cleanly map to cyclic capital-rotation and asset-allocation schedules. We show that the optimal periodic policy is efficiently computable via a reduction to maximum-weight bipartite matchings and captures at least a 1/K fraction of the dynamic optimum. Our main result shows that T^{2/3} is (up to log factors) the minimax rate for the regret (against periodic policies) for adversarial blocking bandits with identical blocking times, and that this rate is achievable by an efficient algorithm.

Vianney Perchet (Crest, ENSAE & Criteo AI Lab)

Last Iterate Convergence for Uncoupled Learning in Zero-Sum Games with Bandit Feedback

Abstract

In this talk, I will introduce the problem of learning in zero-sum game, and especially for the problem of "last-iterate" convergence, unlike the traditional literature that looks at the average convergence (we argue it makes more sense). The interesting property is that the optimal rate is T^{-1/4} which is quite unusual (and unexpected) in this literature. 

https://proceedings.mlr.press/v267/fiegel25a.html

Yang Cai (Yale University)

TBA

Yurong Chen (INRIA Paris) 

Learning a Stackelberg Leader's Incentives from Optimal Commitments

Abstract

Stackelberg equilibria, as functions of the players' payoffs, can inversely reveal information about the players' incentives. In this paper, we study to what extent one can learn about the leader's incentives by actively querying the leader's optimal commitments against strategically designed followers. We show that, by using polynomially many queries and operations, one can learn a payoff function that is strategically equivalent to the leader's, in the sense that: 1) it preserves the leader's preference over almost all strategy profiles; and 2) it preserves the set of all possible (strong) Stackelberg equilibria the leader may engage in, considering all possible follower types. As an application, we show that the information acquired by our algorithm is sufficient for a follower to induce the best possible Stackelberg equilibrium by imitating a different follower type. To the best of our knowledge, we are the first to demonstrate that this is possible without knowing the leader's payoffs beforehand. Due caution is necessary when one intends to utilize the power of optimal commitment. This is a joint work with Xiaotie Deng (Peking University), Jiarui Gan (Uiversity of Oxford), and Yuhao Li (Columbia University).

Important Dates (All times are 11:59 PM AoE)

• Submission Deadline: 31/5/2026

• Notification of Acceptance: 5/6/2026 9/6/2026

• Workshop Date:  July 6, 2025

OLE 2026 aims to provide a venue for researchers to explore and discuss recent trends in topics at the intersection of online learning and economics.  We welcome submissions that explore this space along various directions, including (but not limited) to:

• Learning in repeated auctions

• Learning in mechanism/contract/information design

• Learning in markets

• No-regret learning and convergence to equilibria

• (Online) Calibration

Submission Guidelines

• Submission Platform: https://ole2026.hotcrp.com/

• The preferred format is a 2-page abstract. Longer submissions are welcome, but only the first two pages are guaranteed to be reviewed.

• Submissions will be evaluated based on relevance to the workshop, academic quality, and potential impact.

This is a non-archival workshop. We encourage the submission of work that has been recently published, is under review, or is in progress. Submissions need not be anonymized and authors are encouraged to point to extended versions of their submissions available on public repositories. 

Presentation Format

Authors of accepted submissions will be invited to present their work during a poster session at the workshop. 

• Failure Modes in AI Retraining Dynamics (K. Banihashem, N. Collina, N. Immorlica, B. Lucier, A. Slivkins)

• No-Regret Online Autobidding in Non-Truthful Auctions with ROI and Budget Constraints (Y. Deng, Y. Li, W. Tang, H. Zhang)

• Learning vs. Optimizing Bidders in Budgeted Auctions (G. Fikioris, B. Sivan, E. Tardos)

• On the Impossibility of Information-Value-Free Learning Dynamics: Equilibrium Convergence and Algorithmic Collusion (J. Hartline, C. Wang, C. Zhang)

• MenuNet: A Strategy-Proof Neural Mechanism for Matching Markets (Z. Sun)

• Bandit Social Learning with Exploration Episodes (K. Banihashem, N. Collina, A. Slivkins)

• Learn to Match: Two-Sided Matching with Temporally Extended Feedback (H. Zong, Y. Liang, B. Zhou, N. Jaques)

• Partner Choice in Low-Information Social Dilemmas (S. Roesch, Y. Du, O. Rodrigues, S. Leonardos)

• Learning in Bayesian Stackelberg Games With Unknown Follower’s Types (F. Bacchiocchi, M. Bollini, M. Castiglioni, A. Marchesi, S. Coutts)

• Equilibrium with Internal Transfers (M. Liu, G. Farina, A. Ozdaglar)

• Blackwell Approachability and Gradient Equilibrium are Equivalent (B. W. Lee, N. Haghtalab, M. I. Jordan, R. J. Tibshirani)

• Ex-post equilibria (F. Giordano, J. Grand-Clément, C. Kroer)

• Scale-Invariant Regret Matching and Online Learning with Optimal Convergence: Bridging Theory and Practice in Zero-Sum Games (B. H. Zhang, I. Anagnostides, T. Sandholm)

• Online Learning and Equilibrium Computation with Ranking Feedback (M. Liu, Y. Chen, Z. Fan, G. Farina, A. E. Ozdaglar, K. Zhang)

• Smoothing the Cliff: Incentive-Compatible Priority Allocation via Randomized Mechanisms (T. Lin, S. Yu, H. Zhang)

• Multi-agent Adaptive Mechanism Design (Q. Han, D. Simchi-Levi, R. Tan, Z. Zhao)

• Searching for Optimal Prices in Two-Sided Markets (Y. Feng, M. Ma, B. Peng, Z. Wan)

• Online Learning via Offline Greedy Algorithms: Applications in Market Design and Optimization (R. Niazadeh, N. Golrezaei, J. Wang, F. Susan, A. Badanidiyuru)

• Toward Simultaneously Optimal Regret in U-Calibration (R. Frongillo, H. Luo, N. Mehta, J. Schneider)

• Contracting the misguided agent (J. Tłuczek, E. Yılmaz, V. Villin, C. Dimitrakakis)

• Understanding Strategic Platform Entry and Seller Exploration: A Stackelberg Model (G. Seo, X. Wang, D. C. Parkes)

• When Leaderboards Stop Search: Feedback Precision and Costly Exploration (K. Chen, Y. Lin)

• Swap Regret Minimization Through Response-Based Approachability (I. Anagnostides, G. Farina, M. Fishelson, H. Luo, J. Schneider)

• Learning a Game by Paying the Agents (B. H. Zhang, T. Lin, Y. Chen, T. Sandholm)

• Consumer Search and Social Learning in Agentic Markets (B. Lucier, N. Immorlica, M. Mobius, A. Slivkins, D. Goldstein, J. Hofman, S. Jaffe, D. Rothschild)

• Do Not Trust the Auctioneer: Learning to Bid in Feedback-Manipulated Auctions (L. Foscari, M. Tullii, V. Perchet)

• Learning to Bargain: Last-Iterate Convergence of Follow-the-Regularized-Leader in Games with a Discontinuity (S. Kamp, R. Liebman, B. Fish)

• Dynamic Pricing and Advertising with Demand Learning (S. Agrawal, Y. Feng, W. Tang)

• Contextual Search in Principal-Agent Games: The Curse of Degeneracy (Y. Feng, M. Ma, B. Peng, Z. Wan)

• Robust Learning with Private Information (K. Okumura)

• Profit Maximization in Bilateral Trade against a Smooth Adversary (S. Di Gregorio, P. Duetting, F. Fusco, C. Schwiegelshohn)

ole.ec.2026@gmail.com