IBM Watson for Financial Services Patricia Roszell Business

Agenda § Welcome and Goals/Objectives – Introductions – Client’s Goals – Meeting Objectives § Watson Overview – Key Capabilities – What Makes a Good Watson Use Case – Use Case Critical Success Factors § Current Watson for FSS Solution Concepts – Common Solution Patterns & Financial Services Applications and Benefits – Demonstration – Healthcare “Diagnosis & Act” – Demonstration – Financial Services “Earnings Quality” § Client Use Case Applicability and Brainstorming – High Value Business Outcomes Targeted for Improvement – Unstructured Content (Research, News, Social Media, Client Notes, e. Mails, Voice Conversations, etc. ) and Potential Training Data – Top Performers to Emulate Across the Enterprise § Next Steps 1 © 2011 IBM Corporation

Agenda § Welcome and Goals/Objectives – Introductions – Client’s Goals – Meeting Objectives § Watson Overview – Key Capabilities – What Makes a Good Watson Use Case – Use Case Critical Success Factors § Current Watson for FSS Solution Concepts – Common Solution Patterns & Financial Services Applications and Benefits – Demonstration – Healthcare “Diagnosis & Act” – Demonstration – Financial Services “Earnings Quality” § Client Use Case Applicability and Brainstorming – High Value Business Outcomes Targeted for Improvement – Unstructured Content (Research, News, Social Media, Client Notes, e. Mails, Voice Conversations, etc. ) and Potential Training Data – Top Performers to Emulate Across the Enterprise § Next Steps 2 © 2011 IBM Corporation

Businesses on a Smarter Planet are “dying of thirst in an ocean of data” 90% of the world’s data was created in the last two years 1 in 2 business leaders don’t have access to data they need 80% of the world’s data today is unstructured 83% of CIO’s cited BI and analytics as part of their visionary plan 3 20% is the amount of available data traditional systems leverages 5. 4 X more likely that top performers use business analytics © 2011 IBM Corporation Source: Giga. OM, Software Group, IBM Institute for Business Value"

Financial services firms are beset with some of the most complex information challenges we collectively face § Reuters publishes the equivalent of 9, 000 pages of financial news every day 1 § Five new research documents come out of Wall Street every minute 1 § Asset managers receive up to 1, 000 e-mails daily 1 “The sheer amount of newsflow has made it difficult for people to take positions confidently” Sources: 1 - www. financial-domain. info/integrating-qualitative-and-quantitative-information/ 2 - IBM Client experience with For. Ex traders 4 3 – Derived from NYSE data -Adam Margolis, Citi trader © 2011 IBM Corporation

Today’s business challenges are causing organizations to rethink what it will take to get ahead tomorrow Emerging IT Traditional IT § Structured data (local) § Deterministic Applications § Search Oriented § Small Data § Machine Language 5 § Structured & unstructured (global) § Probabilistic Applications § Discovery Oriented § Small and Big Data § Natural Language © 2011 IBM Corporation

Informed Decision Making: Search vs. Expert Q&A Decision Maker Has Question Search Engine Distills to 2 -3 Keywords Finds Documents containing Keywords Reads Documents, Finds Answers Delivers Documents based on Popularity Finds & Analyzes Evidence Expert Decision Maker Understands Question Asks NL Question Produces Possible Answers & Evidence Considers Answer & Evidence Analyzes Evidence, Computes Confidence Delivers Response, Evidence & Confidence 6 © 2011 IBM Corporation

IBM Watson brings together a set of transformational technologies to drive optimized outcomes Understands natural language and human speech Adapts and Learns from user selections and responses 7 Generates and evaluates hypothesis for better outcomes …built on a massively parallel probabilistic evidence-based architecture © 2011 IBM Corporation

Why is it so hard for computers to understand humans Structured Data Physicist Where was Einstein born? Birth Place A. Einstein Ulm N. Bohr Copenhagen M. Curie Warsaw Source: Excel File, Database, etc. Person Welch ran this? IBM J. Welch GE W. Gates Microsoft “One day, from among his city views of Ulm, Otto chose a water color to send to Albert Einstein as a remembrance of Einstein´s birthplace” Source: http: //www. schaeffenacker-ulm. de/en/otto. html Organization L. Gerstner Unstructured Data Source: Excel File, Database, etc. “If leadership is an art then surely Jack Welch has proved himself a master painter during his tenure at GE” Source: Jack Welch and the GE Way, Robert Slater Source: IBM Research 8 © 2011 IBM Corporation

Watson: Massively Parallel Probabilistic Evidence-Based Architecture 3 Machine 2 Hypothesis Generation & Learning Confidence scoring Learned Models help combine and weigh the Evidence Sources Question Answer Sources Primary Search Candidate Answer 100’s Possible Generation Balance & Combine Models Deep Evidence Retrieval 100, 000’s Scores from Scoring many Deep Analysis 1000’s of Answer Scoring Pieces of Evidence Algorithms Models Models Answers Multiple Interpretations Question & Topic Analysis 1 9 Natural Language Processing 100’s sources Question Decomposition Hypothesis Generation Hypothesis and Evidence Scoring . . . Synthesis Final Confidence Merging & Ranking Answer & Confidence © 2011 IBM Corporation

Deep. QA: Incremental Progress in Answering Precision on the Jeopardy Challenge: 6/2007 -11/2010 IBM Watson Playing in the Winners Cloud 100% 90% v 0. 8 11/10 80% V 0. 7 04/10 Precision 70% v 0. 6 10/09 v 0. 5 05/09 60% v 0. 4 12/08 50% v 0. 3 08/08 v 0. 2 05/08 40% v 0. 1 12/07 30% 20% 10% Baseline 12/06 0% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% % Answered 10 © 2011 IBM Corporation

What Makes a Good Watson Use Case § Would research across a wider variety of data improve decision making? § What is the value of a small set of best answers? § How important is knowing the source and confidence of the answers? § Is there a pool of historically known best answers? 11 © 2011 IBM Corporation

Use Case Critical Success Factors § Does the data exist? (Can it be extracted for Watson? ) § Does training data exist or can it be created? § What is the cost to pursue? (Client and IBM) § Is the outcome defined? (Can you measure) § Is the project credible? (Will it improve results? ) 12 © 2011 IBM Corporation

Agenda § Welcome and Goals/Objectives – Introductions – Client’s Goals – Meeting Objectives § Watson Overview – Key Capabilities – What Makes a Good Watson Use Case – Use Case Critical Success Factors § Current Watson for FSS Solution Concepts – Common Solution Patterns & Financial Services Applications and Benefits – Demonstration – Healthcare “Diagnosis & Act” – Demonstration – Financial Services “Earnings Quality” § Client Use Case Applicability and Brainstorming – High Value Business Outcomes Targeted for Improvement – Unstructured Content (Research, News, Social Media, Client Notes, e. Mails, Voice Conversations, etc. ) and Potential Training Data – Top Performers to Emulate Across the Enterprise § Next Steps 13 © 2011 IBM Corporation

From battling humans at Jeopardy! to transforming how business thinks, acts, and operates Healthcare Diagnostic/treatment assistance, evidencedbased insights, collaborative medicine Financial Services Investment and retirement planning, institutional trading and decision support Contact Center Government Call center and tech support services, enterprise knowledge management, consumer insight Public safety, improved information sharing, security IBM Watson has the capabilities to address grand business and societal challenges 14 © 2011 IBM Corporation

IBM is putting Watson technologies to work in Financial Services Quality of Earnings Institutional Credit Risk Financial Planning Challenge Evaluating earnings is complex, time-consuming and often incomplete Credit provisioning decisions and bond investments are often made without a complete assessment Financial Advisors are challenged to provide personalized advice to clients Solution Leverage unstructured data, evaluate hypothesis, respond with confidence, and learn as it goes – improving valuations Create a single view of the bond for a complete risk profile drawing numerous sources of unstructured information Aggregate and analyze bank and social data to and formulates "best" advice options to investment goals Outcomes Insights into earnings can enhance investor confidence improving the return on investments 16 Outcomes Improved credit decisions and bond investments generate significant returns Outcomes Informed advice and personalized actions improve customer loyalty & revenue © 2011 IBM Corporation

Watson Earnings Quality Prototype 17 Shared Under NDA – Do Not Redistribute © 2011

There are many factors that effect earnings quality Lease Obligations Inventory Turns Tax Acctg. Cash Reserves Off-balance Sheet Purchasing Power Customer Factors Interest Rates Consumer Sentiment Demographics Financial Factors Exchange Rates Product Pipeline Firm Operation Factors Foreign Invest. Industry Factors M&A Activity Litigation Customer Churn Competition M&A Activity New Entrants Product Recalls Gov’t Policy Economic Factors Earnings Quality Pension Plan Funding 18 Labor Rates Regulation Patents / IP M&A Activity Growth Rates © 2011 IBM Corporation

Scoring Earnings Quality Earnings Growth Quantitative Qualitative 19 Earnings Sustainability Time Series Sales Trend Operating Margin (OPM) vs. Asset Turnover (ATO) Change in Receivables minus Change in Sales Excess Cash Margin (ECM) – Operating Earnings – Cash from Operations Change in Inventory minus Change in Sales Accuracy of Earnings Forecasts Investment in Go-To-Market One time revenue Investment in Infrastructure One time expense Investment in Human Capital Leadership / Executive Turnover © 2011 IBM Corporation

Earnings Quality Prototype “… initial results for CETP inhibitors look promising…. . ” An investor reads an article about Merck’s research into a new class of cholesterol reducing drugs “What companies are researching CETP inhibitors? ” “Merck, Pfizer, Eli Lily” The broker uses Watson to learn more about CETP drug 20 market and potential “Merck seems like a good buy, CETP could be a game changer for cardio health…” The investor contacts her broker to ask for a recommendation on purchasing Merck. “What is CETP? ” “CETP stands for Cholesteryl ester transfer protein. It is also known as plasma lipid…. ” The broker uses Watson to learn more about CETP and the potential investment opportunity. Why Pfizer's torcetrapib failure might not thwart other CETP inhibitors” – From Outsourcing Pharma. “Which company is farthest along in cliincal trials for CETP drugs? ” “Eli Lily is is Stage 3 trials as of June 1, 2011. ” Watson’s evidence profile shows the broker that Pfizer has failed so the broker cuts The broker further discovers that Eli Lily, not Merck, is closer to market © 2011 CETP for IBM Corporation

Putting the proper pieces together at point of impact makes for better bottom-line decisions Client requests that trader evaluate Pharma companies with cholesterylester transfer protein (CETP) drugs in pipeline Scenario 21 © 2011 IBM Corporation

Putting the proper pieces together at point of impact makes for better bottom-line decisions TP g Dru ies d Stu Trader explains that CETP studies are just one factor in earnings quality § CETP Drug Studies § CETP News § Regulatory Environment CE Market Data Earnings Quality Acme Pharma, Inc. Pharma United Pharma USA CETP Drug Studies 22 Top Confidence: Pharma Inc. § Acme: Stage 2 failure; start new stage 1 § Pharma Inc: Started Phase 3 trial; 7 completed studies § Pharma United: Phase 2 trials started; 3 completed studies; recruiting for 4 th § Pharma USA: Phase 1 trials complete © 2011 IBM Corporation

Putting the proper pieces together at point of impact makes for better bottom-line decisions § CETP Drug Studies § CETP News § Regulatory Environment Earnings Sustainability § § § Operating Margin Excess Cash Margin Accuracy of Earnings Forecast One Time Revenue One Time Expenses Leadership/Exec Turnover s nse xpe e E tudies Tim S One Drug TP CE Trader knows that earnings sustainability may yield insight into long term stability Market Data Earnings Quality Acme Pharma, Inc. Pharma United Pharma USA One Time Expenses Top Confidence: Pharma USA § Acme: Class action suit is going away which should eliminate one-time expense § Pharma Inc: Setting aside fund for patent infringement claim 23 © 2011 IBM Corporation

Putting the proper pieces together at point of impact makes for better bottom-line decisions Client has expressed interest in learning more about earnings growth impact on company upside § CETP Drug Studies § CETP News § Regulatory Environment Earnings Sustainability § § § Operating Margin Excess Cash Margin Accuracy of Earnings Forecast One Time Revenue One Time Expenses Leadership/Exec Turnover Earnings Growth § Time Series Sales Trend § Change in Receivables minus Change in Sales § Change in Inventory minus Change in Sales § Human Capital 24 ital Cap es an s Hum xpen e E tudies Tim S One Drug TP CE Market Data Earnings Quality Acme Pharma, Inc. Pharma United Pharma USA Top Confidence: Acme § Acme: 260 jobs listed on their website, up 30 from last quarter § Pharma United: Minor layoffs § Pharma , USA: Major layoffs © 2011 IBM Corporation

Putting the proper pieces together at point of impact makes for better bottom-line decisions Infrastructure changes factor into both current cash flows and production § CETP Drug Studies § CETP News § Regulatory Environment Earnings Sustainability § § § Operating Margin Excess Cash Margin Accuracy of Earnings Forecast One Time Revenue One Time Expenses Leadership/Exec Turnover Earnings Growth Infrastructure § Time Series Sales Trend § Change in Receivables minus Change in Sales § Change in Inventory minus Change in Sales § Human Capital § Infrastructure re uctu astr ital Infr Cap es an s Hum xpen e E tudies Tim S One Drug TP CE Market Data Earnings Quality Acme Pharma, Inc. Pharma United Pharma, USA Top Confidence: Acme § Acme: New plant opening § Pharma United: Infrastructure outsourcing 25 © 2011 IBM Corporation

Putting the proper pieces together at point of impact makes for better bottom-line decisions Go to market models can materially affect downstream revenues § CETP Drug Studies § CETP News § Regulatory Environment Earnings Sustainability § § § Operating Margin Excess Cash Margin Accuracy of Earnings Forecast One Time Revenue One Time Expenses Leadership/Exec Turnover Earnings Growth Go To Market 26 § Time Series Sales Trend § Change in Receivables minus Change in Sales § Change in Inventory minus Change in Sales § Human Capital § Infrastructure § Go To Market t rke Ma To cture Go u astr ital Infr Cap es an s Hum xpen e E tudies Tim S One Drug TP CE Market Data Earnings Quality Acme Pharma, Inc. Pharma United Pharma, USA Top Confidence: Acme § Pharma Inc: Partnership with distributor dissolved § Pharma USA: Alliance with major HMO © 2011 IBM Corporation

Watson EQ Prototype – A Phased Approach Phase 1 – Focused Prototype Phase 2 – Iterative Expansion All Financial Services More Markets / Instruments Capital Markets Equities Industry Focus Prototype Question Focus Content Focus Use Case: Deep QA on “Earnings Quality” with a targeted industry, a defined set of questions and a specified content domain (see next slide) 27 More Content Sources More Question Types Build upon initial baseline system by adding more markets/companies, questions, and content (including proprietary) © 2011 IBM Corporation

Agenda § Welcome and Goals/Objectives – Introductions – Client’s Goals – Meeting Objectives § Watson Overview – Key Capabilities – What Makes a Good Watson Use Case – Use Case Critical Success Factors § Current Watson for FSS Solution Concepts – Common Solution Patterns & Financial Services Applications and Benefits – Demonstration – Healthcare “Diagnosis & Act” – Demonstration – Financial Services “Earnings Quality” § Client Use Case Applicability and Brainstorming – High Value Business Outcomes Targeted for Improvement – Unstructured Content (Research, News, Social Media, Client Notes, e. Mails, Voice Conversations, etc. ) and Potential Training Data – Top Performers to Emulate Across the Enterprise § Next Steps 29 © 2011 IBM Corporation

Use Case Applicability and Brainstorming § High Value Business Outcomes Targeted for Improvement § Unstructured Content (Research, News, Social Media, Client Notes, e. Mails, Voice Conversations, etc. ) and Potential Training Data § Top performers to emulate across the enterprise 30 © 2011 IBM Corporation

Agenda § Welcome and Goals/Objectives – Introductions – Client’s Goals – Meeting Objectives § Watson Overview – Key Capabilities – What Makes a Good Watson Use Case – Use Case Critical Success Factors § Current Watson for FSS Solution Concepts – Common Solution Patterns & Financial Services Applications and Benefits – Demonstration – Healthcare “Diagnosis & Act” – Demonstration – Financial Services “Earnings Quality” § Client Use Case Applicability and Brainstorming – High Value Business Outcomes Targeted for Improvement – Use Case Business Processes and Workflows – Unstructured Content (Research, News, Social Media, Client Notes, e. Mails, Voice Conversations, etc. ) and Potential Training Data § Next Steps 31 © 2011 IBM Corporation