2015 Informs Annual Meeting

SD28

INFORMS Philadelphia – 2015

3 - Engineering Sustainable Complex Coevolutionary Agricultural Systems

4 - A Faster Algorithm for Computing Prices in Core-selecting Combinatorial Auctions Benjamin Lubin, Asst Professor, Boston University, 595 Commonwealth Ave, 621A, Boston, Ma, 02215, United States of America, blubin@bu.edu, Benedikt Bunz, Sven Seuken We present a new, faster algorithm for the computationally hard problem of pricing core-selecting combinatorial auctions. First, we provide an alternative definition of the core using weakly stronger constraints. Using these, we offer two new algorithmic techniques that 1) exploit separabililty in allocative conflicts between bidders, and 2) leverage non-optimal solutions. Using large auction instances we show that our algorithm is between 2 and 4 times faster than the current state of the art.

Dr Alejandro N. Martinez-Garcia, Professor, Instituto Tecnologico del Valle de Morelia-Tecnologico Nacional de Mexico, km 6.5 Carretera Morelia-Salamanca s/n, Col. Los Angeles, Morelia, 58100, Mexico, alejandro.martinez.garcia@gmail.com Achieving sustainability (including food security) under the dynamic conditions of climatic change, increasing human population, and poverty reduction, while preserving the ability of ecosystems to provide the services on which humanity depends, implies the need for solving multi-objective optimization problems, under a new paradigm: sustainable complex coevolutionary systems engineering. 4 - A Multi-objective Mathematical Programming Analysis of Forest Carbon Management Midhun Mohan (Mickey), Graduate Student Researcher, North Carolina State University, 15404 Bragaw Hall, Raleigh, NC, 27607, United States of America, mmohan2@ncsu.edu, Henrique Scolforo, Jean Chung, Juan Posse, Tiantian Shen, Bruno Kanieski, Joseph Roise, Glenn Catts, Kevin Harnish This study analyzes the valuation and production possibilities on a working forest using Multi-objective programming, Woodstock, Timber NPV, and Carbon Storage and Sequestration, and present a forest management model for optimizing Net Present Value (NPV) and carbon sequestration at Hofmann forest. SD28 28-Room 405, Marriott Advances in Auction Theory Cluster: Auctions Invited Session Chair: Benjamin Lubin, Asst. Professor, Boston University, 595 Commonwealth Ave, 621A, Boston, Ma, 02215, United States of America, blubin@bu.edu 1 - Are you going to do that? Contingent-payment Mechanisms to Improve Coordination David C. Parkes, Professor, Harvard University, Cambridge, MA, United States of America, parkes@eecs.harvard.edu, Hongyao Ma, Reshef Meir, James Zou We consider coordination problems, such as allocating the right to use a shared sports facility or picking the time of a meeting. Outcomes are designated as either good or bad (is the facility used, will people show up?), and the goal is to attain good outcomes. Reports in period zero about agents’ uncertain values are used to design a choice set for agents in period one, defining also payments that depend on agents’ actions (e.g., using the facility.) 2 - Efficient Interdependent Value Combinatorial Auctions with Single Minded Bidders Valentin Robu, Assistant Professor, Heriot-Watt University, Edinburgh, School Eng. & Physical Sciences, EM3.15, Riccarton Campus, Edinburgh, EH144AS, United Kingdom, V.Robu@hw.ac.uk, David C. Parkes, Takayuki Ito, Nicholas R. Jennings We study the design of efficient auctions where bidders have interdependent values, that depend on signals of other bidders. In particular, we consider a contingent bid model in which bidders may explicitly condition the value of their bids on the bids submitted by others. We derive constraints which allows the efficient second price, fixed point auction to be implemented in single-minded CAs, and present an alternative mechanism for cases in which the required single crossing condition fails. 3 - New Core-selecting Payment Rules with Better Fairness and Incentive Properties Sven Seuken, Assistant Professor Of Computation And Economics, University of Zurich, Binzmuhlestrasse 14, Zurich, ZH, 8050, Switzerland, sven.seuken@gmail.com, Benjamin Lubin, Benedikt Bönz We introduce four “Small” rules, which are new core-selecting payment rules for combinatorial auctions. Via a Bayes-Nash equilibrium analysis, we first show for a domain with 2 goods and 3 bidders, that one of our rules outperforms the state- of-the-art Quadratic rule along all dimensions (efficiency, incentives, fairness, and revenue). We then use a computational approach to evaluate 85 different rules in a setting with 25 goods and 10 bidders, and show that our new rules still perform best.

SD29 29-Room 406, Marriott Baseball Analytics Sponsor: Analytics Sponsored Session

Chair: Sean Barnes, University of Maryland, 4352 Van Munching Hall, University of Maryland, College Park, MD, 20742, United States of America, sbarnes@rhsmith.umd.edu Co-Chair: Margret Bjarnadottir, Assistant Professor of Management Science and Statistics, Robert H. Smith School of Business, University of Maryland, 4324 Van Munching Hall, College Park, MD, 20742, United States of America, margret@rhsmith.umd.edu 1 - The Effectiveness of Dynamic Pricing Strategies on Single-game Ticket Revenue in Baseball Joseph (Jiaqi) Xu, The Wharton School, University of Pennsylvania, 3730 Walnut Street, Suite 500, Philadelphia, PA, United States of America, jiaqixu@wharton.upenn.edu, Peter Fader, Senthil Veeraraghavan We develop a comprehensive demand model for single-game tickets that can be used to predict revenue associated with a particular pricing strategy over the course of sport season. We apply the model to actual sales and pricing data from an anonymous MLB franchise during a recent baseball season and evaluate the effectiveness of the dynamic pricing policy. We propose pricing heuristics and find that optimized dynamic pricing policy can improve revenue by 14.3% compared to a flat pricing policy. 2 - Pitch Sequence Complexity and Long-Term Pitcher Performance Joel Bock, Booz Allen Hamilton, 901 15th Street NW, Washington, DC, United States of America, sauerkraut@gmail.com Patterns of a baseball pitcher’s pitch type sequencing can be learned by machine learning models trained on historical data. Individual pitch-wise predictability is connected with broader performance statistics (ERA, FIP) by a regression model that may be used to forecast player performance. Less complexity correlates with higher values of ERA and FIP. This talk outlines the analytical approach and presents results from a study of Major League Baseball pitchers covering three recent seasons. 3 - The Value of Positional Flexibility Timothy Chan, University of Toronto, 5 King’s College Road, Toronto, ON, M5S 3G8, Canada, tcychan@mie.utoronto.ca, Douglas Fearing Drawing from the theory of production flexibility in manufacturing networks, we provide the first optimization-based analysis of the value of positional flexibility (the ability of a player to play multiple positions) for a major league baseball team in the presence of injury risk. Using publicly available data on baseball player performance, we derive novel baseball-related insights that can be generalized to the manufacturing context. 4 - A Bayesian Hierarchical Model for Modeling Called Strikes in Major League Baseball Abraham Wyner, Professor, University of Pennsylvania, The Wharton School, University of Pennsy, 400 JMHH, Philadelphia, Pe, 19104, United States of America, ajw@wharton.upenn.edu, Sameer Deshpande We measure a catcher’s ability to “frame” a pitch. The effect exists, but there remains debate on the effect size. We introduce a systematically constructed, parametric Bayesian hierarchical model for the probability of a called strike. Our model adjusts, accounts borrows strength from data on all participants. By sharing information across all participants we are able to accurately quantify the player’s framing effect on a pitch and translate that effect into runs added across a season.

128