Core Viewpoint - Artificial Intelligence (AI) is rapidly transforming the financial services landscape, with a strong emphasis on the importance of data quality for the success of AI models [3][4][62]. Group 1: Importance of Data in AI - The performance of AI models is entirely dependent on the quality of the data they absorb, as highlighted by LSEG's CEO David Schwimmer [3]. - Financial data is complex, fragmented, and often subject to regulatory constraints, encompassing both structured and unstructured formats [3][4]. - Optimizing financial data for AI requires domain expertise, robust infrastructure, and meticulous governance [3][4]. Group 2: Challenges in Financial AI - Up to 85% of financial AI projects fail due to data quality issues, talent shortages, and strategic misalignment [4]. - Gartner predicts that 30% of generative AI projects will be abandoned after the proof-of-concept phase due to poor data quality [4]. Group 3: Data Categories and Optimization Techniques - **Macroeconomic Data**: Includes indicators like CPI, GDP, and unemployment rates, crucial for predictive models and trading signals [9]. - Optimization techniques involve using point-in-time (PIT) and real-time data to avoid biases from historical corrections [11]. - **Pricing Data**: Forms the basis for security valuation, including real-time quotes and historical prices [14]. - Risks include misleading models due to lagged and revised data [15]. - **Reference Data**: Provides descriptive details about securities and entities, essential for filtering trading eligibility and detecting anomalies [20]. - Optimization techniques include creating master mapping tables and tracking data lineage [24]. - **Symbol Mapping**: Involves using identifiers like ISIN and CUSIP to map and stitch datasets together [27]. - Risks include identifier changes due to corporate actions [29]. - **Unstructured Text**: Comprises news, research reports, and records, rich in insights but challenging to process [35]. - Techniques include using natural language processing for summarization and sentiment analysis [38]. - **Company Data**: Includes structured financial data and unstructured disclosures, vital for valuation and ESG analysis [42]. - Risks involve misinformation and misinterpretation [43]. - **Risk Intelligence Data**: Encompasses sanctions, politically exposed persons, and negative news, critical for compliance and fraud detection [49]. - Optimization techniques focus on standardizing names and addresses [51]. - **Analysis**: Used for valuation, hedging, and risk metrics, potentially involving local or cloud-based computing engines [57]. - Techniques include automating anti-money laundering and fraud detection [59]. Group 4: Conclusion on AI Readiness - The success of AI in financial institutions hinges not only on sophisticated algorithms but also on the integrity and readiness of the underlying data [62]. - Optimizing financial data is an ongoing task requiring collaboration among data engineers, domain experts, and AI practitioners [62].