A generative solution for ATM Cash Management Roberto Armenise ∗ , Cosimo Birtolo ∗ , Eugenio Sangianantoni ∗ , and Luigi Troiano † ∗ Poste Italiane S.p.A. – TI - SSI - Centro Ricerca e Sviluppo 80133 Napoli, Italy {armenis5, birtoloc, e.sangianantoni}@posteitaliane.it † University of Sannio – Department of Engineering 82100 Benevento, Italy troiano@unisannio.it Abstract Optimizing cash in automatic teller machines (ATM) is challenging due unpredictability of withdrawals, but prof- itable because of the large number of tellers. Generally ATM cash management and optimization is performed manually, according to corporate policies and personnel experience. A non-optimal cash upload can lead to poor service when cash demand is underestimated and to unnecessary costs when demand is overestimated. Therefore, finding the best match between cash stock and demand becomes crucial to improve. Recently, some authors attempted to optimize the cash by modeling and forecasting the demand. However, the high variance and non-stationarity of the underlying stochastic process can affect reliability of such an approach. In this paper we suggest the application of genetic algorithms as means for searching and generating optimal upload strategies, able at the same time to minimize the daily amount of stocked money and to assure cash dispensing service. Experimentation led at Poste Italiane S.p.A. makes this promising and worth to be further investigated. 1. Introduction Automatic Teller Machines (ATM) are common means to dispense cash. Nowadays ATM are ubiquitous especially in urban areas, exceeding 1.6 million worldwide according to 2007 survey of ATM Industry Association [1]. Cash management and forecasting have been becoming important features for ATM network of next generation [2]. ATMmarketplace.com performed a survey aimed at understanding ATM features as far as implemented by current ATM software solutions and desired by buyers [3]. Respondents were financial institutions worldwide, based in North America (36%), Europe (23%), Asia (19%), Middle East/Africa (12%), Latin America (8%) and Australia (2%). The total number of respondents were 206 in 2009, and 243 in 2010. According to this study, Tab.1 reports the three most desired ATM functionalities. Cash management and forecasting placed 3rd after remote monitoring and cross- platform multivendor ATM software in 2010, moving from 6th place in 2009 ranking. Features 2010 2009 Remote monitoring of the ATM network 49% 45% Multivendor ATM software 31% 22% Cash management and forecasting 28% 22% Envelope-free check deposit 25% 24% One-to-one marketing/purchase gift cards 24% 26% Customer preferences 23% 26% Bulk-note cash deposit 22% 23% Software distribution 22% 28% Support for cash recycling 21% 21% Automated test tools 15% 18% Support for biometrics 13% 13% Other 6% 7% Table 1: The three most desired new features of ATM software (Percentage of financial institutions including the features among their top three priorities) [3]. Poste Italiane S.p.A. is a leading operator of postal ser- vices and an innovative and competitive player in the arena of financial and payment services. It provides the public mail service and it boasts a network of nearly 14,000 post offices and 5,900 ATMs. Stocking cash in ATM entails costs that can be broadly divided in two contributions: financial costs and operational costs [4]. The first are mainly due to unused stock rated by annual passive interests, while time to perform and supervise the task, maintenance, out-of-service and risk of robbery are associated to the second. According to Simutis et al. [5], effective cash management should rely on advanced algorithms able to accurately pre- 349 978-1-4244-7896-5/10/$26.00 c 2010 IEEE