About

Alessandro Arlotto is an Associate Professor of Business Administration, Mathematics, and Statistical Science at Duke University. He holds a primary appointment in the Decision Sciences area of Duke's Fuqua School of Business and secondary appointments in the departments of Mathematics and Statistical Science. Alessandro received his Ph.D. in 2012 from the University of Pennsylvania and joined Duke University in the same year. His research interests encompass probability, optimization, and their applications to business and economics. His work has been published in several journals including the Annals of Applied Probability, Management Science, Mathematics of Operations Research, Operations Research, and Stochastic Processes and their Applications. Alessandro is a recipient of the Faculty Early Career Development (CAREER) award from the National Science Foundation. At Duke, he teaches core courses such as Probability and Statistics in the Daytime and Executive MBA programs, as well as the Quantitative Business Analysis course for the Master in Management Studies. He also teaches the graduate course Stochastic Models.

Research topics

• Computer Science
• Operations research
• Mathematical optimization
• Machine Learning
• Mathematics
• Statistics
• Mathematical economics
• Engineering
• Geometry

Selected publications

• Online Demand Fulfillment Problem with Initial Inventory Placement: A Regret Analysis

  Operations Research · 2026-02-12

  article1st authorCorresponding

  Studying How Inventory Placement Shapes Online Fulfillment Performance A growing share of e-commerce operations involves deciding not only how to fulfill incoming orders, but also where to position inventory beforehand. In “Online Demand Fulfillment Problem with Initial Inventory Placement: A Regret Analysis,” Arlotto, Keskin, and Wei examine how these two decisions interact. The authors introduce a joint regret framework that evaluates the performance of a fulfillment policy together with the initial allocation of inventory across warehouses. Their analysis shows that probabilistic fulfillment inevitably accumulates regret that grows with the length of the planning horizon, regardless of how inventory is initially placed. By contrast, when combined with an appropriate offline inventory placement, the score-based approach achieves a regret bound that remains stable over time and scales only with the size of the system. The study offers insight into the role of initial placement in shaping fulfillment policy performance.

  PublisherDOI

• Ballot Design and Electoral Outcomes: The Role of Candidate Order and Party Affiliation

  ArXiv.org · 2025-07-22

  preprintOpen access1st authorCorresponding

  We use causal inference to study how designing ballots with and without party designations impacts electoral outcomes when partisan voters rely on party-order cues to infer candidate affiliation in races without designations. If the party orders of candidates in races with and without party designations differ, these voters might cast their votes incorrectly. We identify a quasi-randomized natural experiment with contest-level treatment assignment pertaining to North Carolina judicial elections and use double machine learning to accurately capture the magnitude of such incorrectly cast votes. Using precinct-level election and demographic data, we estimate that 11.8% (95% confidence interval: [4.0%, 19.6%]) of democratic partisan voters and 15.4% (95% confidence interval: [7.8%, 23.1%]) of republican partisan voters cast their votes incorrectly due to the difference in party orders. Our results indicate that ballots mixing contests with and without party designations mislead many voters, leading to outcomes that do not reflect true voter preferences. To accurately capture voter intent, such ballot designs should be avoided.

  PublisherOA PDFDOI

• Dynamic Resource Allocation: The Geometry and Robustness of Constant Regret

  Mathematics of Operations Research · 2024 · 2 citations

  • Computer Science
  • Mathematics
  • Mathematical optimization

  We study a family of dynamic resource allocation problems, wherein requests of different types arrive over time and are accepted or rejected. Each request type is characterized by its reward, arrival probability, and resource consumption. An upper bound for the collected reward is given by a linear optimization problem with a random right-hand side. This type of problem, known as packing linear program (LP), is ubiquitous in resource allocation. We provide a detailed characterization of the parametric structure of this packing LP. Relying on this geometric understanding, we revisit and expand on BudgetRatio algorithms that achieve constant regret by resolving this same packing LP in each period and accepting requests scored as sufficiently valuable. We illustrate the benefits of the geometric view in proving that (i) BudgetRatio achieves constant regret relative to the offline (full information) upper bound in the presence of inventory that is (slowly) restocked, and (ii) within explicitly identifiable bounds, the algorithm’s regret is robust to misspecification of the model parameters. This gives bounds for the bandits version of the problem in which the parameters have to be learned. (iii) The algorithm has an equivalent formulation as a generalized bid-price algorithm in which the bid prices can be adaptively and efficiently computed. Our analysis focuses on the evolution of the remaining inventory—in turn of the LP that drives BudgetRatio—as a stochastic process. We prove that it is attracted to sticky regions of the state space in which the online algorithm takes actions consistent with the optimal basis of the offline upper bound, a basis that is revealed only in hindsight at the horizon’s end. Funding: This work was supported by the U.S. Department of Defense [Grant W911NF-20-C-0008].

  PublisherDOI

• Online Demand Fulfillment Problem with Initial Inventory Placement: A Regret Analysis

  SSRN Electronic Journal · 2023 · 1 citations

  1st authorCorresponding
  • Computer Science
  • Computer Science
  • Operations research
  PublisherDOI

• Logarithmic Regret in the Dynamic and Stochastic Knapsack Problem with Equal Rewards

  Stochastic Systems · 2020 · 11 citations

  1st authorCorresponding
  • Computer Science
  • Mathematics
  • Mathematical optimization

  We study a dynamic and stochastic knapsack problem in which a decision maker is sequentially presented with items arriving according to a Bernoulli process over n discrete time periods. Items have equal rewards and independent weights that are drawn from a known nonnegative continuous distribution F. The decision maker seeks to maximize the expected total reward of the items that the decision maker includes in the knapsack while satisfying a capacity constraint and while making terminal decisions as soon as each item weight is revealed. Under mild regularity conditions on the weight distribution F, we prove that the regret—the expected difference between the performance of the best sequential algorithm and that of a prophet who sees all of the weights before making any decision—is, at most, logarithmic in n. Our proof is constructive. We devise a reoptimized heuristic that achieves this regret bound.

  PublisherDOI

• Uniformly Bounded Regret in the Multisecretary Problem

  Stochastic Systems · 2019-08-30 · 82 citations

  articleOpen access1st authorCorresponding

  In the secretary problem of Cayley [Cayley A (1875) Mathematical questions with their solutions. Ed. Times 23:18–19.] and Moser [Moser L (1956) On a problem of Cayley. Scripta Mathematica 22(3/4):289–292.], n nonnegative, independent, random variables with common distribution are sequentially presented to a decision maker who decides when to stop and collect the most recent realization. The goal is to maximize the expected value of the collected element. In the k-choice variant, the decision maker is allowed to make k ≤ n selections to maximize the expected total value of the selected elements. Assuming that the values are drawn from a known distribution with finite support, we prove that the best regret—the expected gap between the optimal online policy and its offline counterpart in which all n values are made visible at time 0—is uniformly bounded in the number of candidates n and the budget k. Our proof is constructive: we develop an adaptive budget-ratio policy that achieves this performance. The policy selects or skips values depending on where the ratio of the residual budget to the remaining time stands relative to multiple thresholds that correspond to middle points of the distribution. We also prove that being adaptive is crucial: in general, the minimal regret among nonadaptive policies grows like the square root of n. The difference is the value of adaptiveness.

  PublisherOA PDFDOI

• Logarithmic regret in the dynamic and stochastic knapsack problem.

  arXiv (Cornell University) · 2018-09-06 · 5 citations

  articleOpen access1st authorCorresponding
  Publisher

• Strategic Open Routing in Service Networks

  Management Science · 2018-05-18 · 14 citations

  article1st authorCorresponding

  We study the behavior of strategic customers in an open-routing service network with multiple stations. When a customer enters the network, she is free to choose the sequence of stations that she visits, with the objective of minimizing her expected total system time. We propose a two-station game with all customers present at the start of service and deterministic service times, and we find that strategic customers “herd,” that is, in equilibrium all customers choose the same route. For unobservable systems, we prove that the game is supermodular, and we then identify a broad class of learning rules—which includes both fictitious play and Cournot best response—that converges to herding in finite time. By combining different theoretical and numerical analyses, we find that the herding behavior is prevalent in many other congested open-routing service networks, including those with arrivals over time, those with stochastic service times, and those with more than two stations. We also find that the system under herding performs very close to the first-best outcome in terms of cumulative system time. The online appendices are available at https://doi.org/10.1287/mnsc.2017.2971 . This paper was accepted by Gad Allon, operations management.

  PublisherDOI

• Logarithmic regret in the dynamic and stochastic knapsack problem with equal rewards

  arXiv (Cornell University) · 2018-09-06 · 2 citations

  preprintOpen access1st authorCorresponding

  We study a dynamic and stochastic knapsack problem in which a decision maker is sequentially presented with items arriving according to a Bernoulli process over $n$ discrete time periods. Items have equal rewards and independent weights that are drawn from a known non-negative continuous distribution $F$. The decision maker seeks to maximize the expected total reward of the items that she includes in the knapsack while satisfying a capacity constraint and while making terminal decisions as soon as each item weight is revealed. Under mild regularity conditions on the weight distribution $F$, we prove that the regret---the expected difference between the performance of the best sequential algorithm and that of a prophet who sees all of the weights before making any decision---is, at most, logarithmic in $n$. Our proof is constructive. We devise a reoptimized heuristic that achieves this regret bound.

  PublisherOA PDFDOI

• An adaptive <i>O</i>(log <i>n</i>)‐optimal policy for the online selection of a monotone subsequence from a random sample

  Random Structures and Algorithms · 2017-08-14 · 6 citations

  articleOpen access1st authorCorresponding

  Given a sequence of n independent random variables with common continuous distribution, we propose a simple adaptive online policy that selects a monotone increasing subsequence. We show that the expected number of monotone increasing selections made by such a policy is within of optimal. Our construction provides a direct and natural way for proving the ‐optimality gap. An earlier proof of the same result made crucial use of a key inequality of Bruss and Delbaen [5] and of de‐Poissonization.

  PublisherOA PDFDOI

Recent grants

• CAREER: The Effects of Centralized and Decentralized Sequential Decisions on System Performance

  NSF · $500k · 2016–2021

Frequent coauthors

• J. Michael Steele

  16 shared

• Yehua Wei

  10 shared

• Xinchang Xie

  Northwestern University

  4 shared

• Irem Nur Keskin

  Duke University

  4 shared

• Noah Gans

  University of Pennsylvania

  4 shared

• Robert W. Chen

  3 shared

• Andrew Frazelle

  The University of Texas at Dallas

  3 shared

• Lawrence A. Shepp

  University of Pennsylvania

  3 shared

Education

• Ph.D.

  University of Pennsylvania

  2012

Awards & honors

• Faculty Early Career Development (CAREER) award from the Nat…