Webinar August 10 2019 1200 Hrs Analytics in

Webinar August 10, 2019 1200 Hrs. Analytics in Banking Vineet Khanna Executive Director – SAS Institute India (Pvt) Ltd. Email: Vineet. Khanna@sas. com

Analytics in Banking • Situation & Trends in Banking industry. • Analytics / AI & ML • A few examples of adoption of Analytics by banks: • • Risk Management Controlling Financial Crimes Contextual Marketing Evolving Areas www. actuariesindia. org

Asia Pacific Banking Source: Mc. Kinsey, Asia Pacific Banking Review 2019 www. actuariesindia. org

Strategic Priorities Improve Productivity Grow Business Digital Fraud Control Risk Management Optimize Capital www. actuariesindia. org

Analytics in Banking • Situation & Trends in Banking industry. • Analytics / AI & ML • A few examples of adoption of Analytics by banks: • • Risk Management Controlling Financial Crimes Contextual Marketing Evolving Areas www. actuariesindia. org

Analytics Trends Data Flat Files Analysis Classic Statistical Methods 1970 Relational databases (Oracle, DB 2), data integration Advanced analytics methods, multistep process flows 1980 Computing Paradigms Monoliths Open / networked systems Technology Platforms Mainframe Minicomputers, internal networks, UNIX, PCs Data warehouses Interactive interfaces, code generation, SQL pushdown, remote submit, business rules 1990 Data virtualization, MDM, unstructured data Grid-based analytics, distributed processing, in-memory processing, text analysis, model management Big Data (Hadoop), data governance, cloud, streams In-database processing, in-stream processing, machine learning, deep learning 2010 2000 Client-server, Internetbased systems In-memory, multicore, virtualization Networks, Linux Mobile, grids In-database, cloud, Internet-of-Things, APIs Cloud-based infrastructure, platforms-as-a-service www. actuariesindia. org

Analyst Forecast "The AI market continued to grow at a steady rate in 2018, and we expect this momentum to carry forward over the forecast period. IDC forecasts the overall market to maintain a steady growth rate annually through 2023, approaching $98. 4 billion in revenue at a CAGR of 28. 5%, " "Investments in analytics and artificial intelligence are driven by the promise, opportunity, and excitement of a new wave of automation that not only drives inefficiency out of processes but also changes how people interact with the digitized world around them and how processes and whole ecosystems change because of automation, " "However, investment in analytics and AI will be moderated by shortage of algorithm training data, outdated legal frameworks, shortage of analytics staff, behavioral biases, and insufficient attention to analytic orientation and data literacy. " Source : IDC Website www. actuariesindia. org

Analytics in Banking • Situation & Trends in Banking industry. • Analytics / AI & ML • A few examples of adoption of Analytics by banks: • • Risk Management Controlling Financial Crimes Contextual Marketing Evolving Areas www. actuariesindia. org

Risk Analytics ENTERPRISE RISK MANAGEMENT DASHBOARDS AND REPORTS Capital Planning, IFRS 9 EST RBS & ICAAP MARKET RISK CREDIT RISK ALM Operational Risk Instrument Valuation Credit Scoring Cash Flow generation EGRC Monitor Risk Measures Validations, Calibration Models – CASA, Options Op. Risk Global Data Counterparty Credit Risk Economic Capital Funds Transfer Pricing Op. Risk Va. R STD & IMM Compliance, STD & IRB Compliance, Liquidity Risk Compliance Management FRTB Portfolio Optimization Cash Flow Optimization Audit Management Risk Based Limits Stress Testing & Scenario Analysis RISK INFRASTRUCTURE Recalibration Auditability Re Runs Comparatives Data Governance Model Risk Integrations Treasury Core Banking Loan Origination System MARKET DATA www. actuariesindia. org

Market Risk Valuations Risk Reporting • Bootstrapping • Valuation – FIMMDA / FEDAI • Derivatives Greeks • Cash flow • Pre/Post Trade Processing • Standardized Approach • IMA • Risk Factor Modelling • Stress Testing • CVA / SA CCR • Sensitives • FRTB • Data Quality • Disallowance • Back testing • RBI Regulatory • Limit Management • Internal MIS • In Memory Reporting Performance Management • Performance Analysis – Sharpe, Jensen, Sortino etc. . • Portfolio Optimization Efficient Frontier • Portfolio Attribution www. actuariesindia. org

Market Risk www. actuariesindia. org

IFRS 9 Data Collection New Information Individual Account Level Forecasts / Historical Segmentation Individual Asset Level = Massive Amount Data = More Granular Data Forward Looking Calculations Financial Impact Increased Measurement complexity Additional Data Collection More Risk Models = New Analytical Models Governance Documentation Governance Change Control Regulatory Capital forecast Model Management = New Control Audit Preparation Framework One of the most challenging areas of an ECL implementation will be aligning the banks interpretation with what is deemed acceptable by the auditor and regulator. = Risk and Finance Integration www. actuariesindia. org

IFRS 9 www. actuariesindia. org

Financial Crime Compliance AML Customer Risk Rating Early Warning Signals Threshold Optimization Internet Banking Mobile Banking ATM Application Frauds Trade Finance Branch Banking Cards UPI Internal Fraud Office Accounts Fraud Business Modules Fraud and Financial Crimes Management Platform Enterprise Data Orchestration Advanced Analytics & Machine Learning Alert Triage Case Management Ops and Management Reporting & Workflows www. actuariesindia. org

Anti Money Laundering www. actuariesindia. org

Fraud Business Rules Anomaly Detection Predictive Models / Advanced Analytics Social Network Analysis Text Mining www. actuariesindia. org

AML & Fraud www. actuariesindia. org

AML & Fraud Application Area Business Value Name Entity Context Recognition Detecting the topics and subject of interest from unstructured text (example : SWIFT) Individual Profile Violation Peer Profile Violation Anomaly Detection in transaction behavior as an individual or peer group (ex: Quick Service Restaurants) New Scenario Detection Detect new modus operandi or change in existing modus operandi of financial crime Suppression Rules Development Financial Crime Prediction Alert Investigation Optimization Change existing thresholds settings to reduce False Positives Identify look alike profiles for undetected ones Rescore the alerts to prioritize investigation based on revised score AI / ML Technique Bayesian Networks Boolean Rules Decision Trees Factorization Machines Frequent Item Set Mining Gradient Boosting K Nearest Neighbor Image Processing Market Basket Analysis Moving Windows PCA Network Analytics/Community Detection Neural Networks / Deep Learning Random Forest Robust PCA Support Vector Data Description (SVDD) Support Vector Machines Text Mining Variable Clustering www. actuariesindia. org

Contextual Marketing Hey Eve! Your current balance is now € 67, but we have good news. Text “Y” to activate a € 200 credit extension on your account 3: 29 pm Dress Track customer behaviour and look for specific patterns (i. e. 3 transactions within 30 minutes resulting in low balance) 3: 46 pm Jeans 3: 57 pm Scarf Eve is shopping, her balance is close to zero and she is eligible for contact 3: 57 pm Offer <200 Ms Apply business logic to determine best action Calculate personalized offer parameters Deliver assistance, guidance or offer Response capture & fulfillment Shopping = • 3 txns in <30 mins • same debit card www. actuariesindia. org

Behind the Scenes Check for Card w/Headroom Check for Deposit Account Encourage Card Use Pre-Approved Overdraft? Recommend Transfer Advise Mobile App or Branch Offer Overdraft Run Credit Risk Check Evaluate Potential Offers Warning! No Offer New Overdraft Offer Personal Loan Offer Credit Card www. actuariesindia. org

Cyber Analytics Information Retrieval Meets Intrusion Detection Spatial Temporal Analysis System bases identification of security incidents and system failures on event log messages Text analytics based Parsing, topic generation and term frequency and inverse frequency calculated TF-IDF weighting calculation to determine whether a new log message deviates from the norm. Less frequently occurring terms are given a higher weight commonly occurring log messages have a lower weight than do those with infrequently occurring elements A list of terms and corresponding IDF weights can be stored as a hash, which generally yields fast results for searching Rarity Detection www. actuariesindia. org

Key Takeaways Analytics is a rapidly changing & fast growing area Banks are adopting Analytics to solve business problems Demand for data scientists is quite high An Actuary brings a lot more to the table It’s a win-win situation for Banks and Actuaries www. actuariesindia. org