Lottery Data Analytics Data AnalysisInvestigations October 2013 Data
Lottery Data Analytics Data Analysis/Investigations October 2013
Data Analysis prior to 2009 • Investigations with no data analysis to support claims • Difficult to obtain data due to strict internal controls on access • Data was available but not organized or in one central program or system • 30 day limit on availability of lottery transactions (purchases/validations) • Customers were responsible to know their prize and ticket in hand held a lot of weight in prize payment Lottery Data Analytics – October 2013 2
OLG Systems and Tools Lottery Data Analytics – October 2013 3
Organizational Changes • Fifth Estate program focused on OLG • Ombudsman’s Report • Media and public interest in suspicious prize claims by insiders Lottery Data Analytics – October 2013 • OLG changed to increase: – – – Integrity Risk Management Customer Focus Controls Data Analysis 4
Data Analysis – A New Team • Two teams were created to help modernize our systems to support data analysis – Technology: developing the tools and programs to assist in analysis – Analysis: extract the data and analyze the information to support claims and identify retailer theft, fraud or dishonestly Lottery Data Analytics – October 2013 5
Current Capabilities • Selection Searches • Retailer comparison reports • Lottery transactions (purchase/validations) back to March 1999 • Prize claims processed through the prize centre (customer’s previous wins) Lottery Data Analytics – October 2013 6
Developing Customer Profiles • Following these tickets we can develop unique play patterns such as: – unique spend could be the dollar value of total purchases, – whether or not the customer plays selection or quick pick, – whether or not the customer plays ENCORE and the number of ENCORE played – what lottery products they play/mix of lottery products they play – the unique locations that the customer purchases/validates their tickets, – the cities that the customer purchases/validates ticket and – the days of the week/time of the day that they are purchasing/validating their tickets Lottery Data Analytics – October 2013 7
Profiling Selections • One of the first charges laid by the OPP on lottery theft was a claim from a retailer. – The winning ticket was a selection (customer picked their own numbers). • OLG reviewed lottery transactions following the selection back from 2001 to 2008 • Develop a profile that included: – – Location/city that the customer played at Day and time of purchases/validations Identify slight change in the numbers selected Changes to customer spend • This was OLG’s first player profile Lottery Data Analytics – October 2013 8
Chung Case • Profiling was done on both the customer and on the retailer behaviour • What we learned about the retailer: – Anomalous Retail Behaviour – Validated these tickets together on certain days of the week • What we learned about the customer – – Played on certain days and time Played between two cities Specific spend on specific product This told us the customer lived in one city and worked in another – Potential to be a group play Lottery Data Analytics – October 2013 9
Profiling Today’s Winners • • • Ability to profile the customer and develop questions where we already know the answers For one of OLG’s latest winners we were able to build a robust profile of the customer including – where we believed he lived – where and when he purchased lottery tickets – how he validated tickets – Lottery products played and spend – how long he has been playing the lottery – where he vacationed this summer OLG was also able to identify that the winning ticket was not likely a group based on the numbers selected. Lottery Data Analytics – October 2013 Product LMAX Selection 5 7 11 19 24 38 49 1 4 5 15 20 33 49 4 8 15 16 23 42 49 4 19 20 22 33 38 49 3 5 10 11 33 38 49 LMAX 1 11 18 23 33 38 49 1 7 8 23 38 39 49 LMAX 2 3 5 10 19 33 49 4 7 10 11 33 38 49 LMAX 4 11 19 24 33 38 49 10
- Slides: 10