Q2B 2020 | BBVA Quantum-Ready journey, towards a Corporate Quantum Mindset

Quantum Credit Journey: Building an Internal Quantum Mindset

Introduction to Quantum Credit Journey

• The speaker introduces the topic of their quantum credit journey, emphasizing the collaboration between the research team and other departments to foster an internal quantum mindset.

• Initial focus was on understanding quantum computing's potential, with a scouting phase that assessed state-of-the-art technologies.

Discovery Phase and Problem Identification

• After determining a high probability (70-80%) for impactful outcomes from quantum computing within three years, they moved into the discovery phase.

• Formation of a dedicated research team began, focusing on identifying key problems to solve while considering their impact on profit and loss (P&L).

Collaboration and Integration

• The organization is now in phase three of its quantum journey, aiming to understand end-to-end processes for implementing solutions in production.

• A collaborative structure was established, referred to as an "onion" or "molecule" model, integrating various business units and engineering teams.

Agnostic Approach to Hardware

• Emphasis on being agnostic regarding hardware choices due to uncertainty about future winners in technology; focus remains on software and models instead.

• Partnerships with research centers aim at exploring new financial applications using quantum models without bias towards specific backends.

Understanding Exponential Problems in Finance

Introduction by Scholastico Sanchez

• Scholastico Sanchez introduces himself as a mathematician who transitioned into finance following significant market events like the bankruptcy of Long-Term Capital Management.

Key Questions Addressed

• Three critical questions guide their research:

• What problem are we dealing with?

• How do we currently address these problems?

• What capabilities can we develop in quantum computing?

Focus on Exponential Problems

• Discussion revolves around exponential problems in finance, specifically referencing MP-hard problems related to pricing options under varying market conditions.

Market Dynamics Explained

• Explanation of how price movements affect volatility; practitioners recognize that rising prices typically correlate with declining volatility and vice versa.

Quantum Computing: Understanding Its Advantages

Introduction to Quantum Computing

• The speaker discusses the limitations of classical computing when faced with complex calculations involving numerous factors, emphasizing that quantum computing can handle these challenges more efficiently.

• Key properties of quantum computing are introduced: superposition and entanglement, which allow for advanced mathematical manipulations through tensor products of vector spaces.

Applications in Business Lines

• The organization has explored various business lines including asset management and corporate investment banking, identifying high-volume problems suitable for quantum solutions.

• A scientific deductive method is applied in three phases: data collection, hypothesis formulation (quantum advantage in optimization and Monte Carlo simulations), and proof of concept development.

Proof of Concept in Optimization

• Six proofs of concept have been completed focusing on optimization and Monte Carlo simulations, filtering over 10 identified problems based on feasibility and impact.

• One example discussed is dynamic portfolio optimization, which considers maximizing returns while minimizing risks, transaction costs, and fees across a large dataset.

Results from Portfolio Optimization

• A significant achievement includes optimizing a portfolio with 52 assets over eight years of daily data using quantum methods, yielding impressive results compared to classical computing standards.

• The results indicate a Sharpe ratio of 12 for one optimized portfolio, suggesting substantial potential returns against minimal risk tolerance.

Further Developments in Corporate Investment Banking

• Another proof of concept focuses on valuing financial products with credit value adjustments (CVA), developed collaboratively with Zapata Computing.

• This project addresses the complexities involved in pricing derivatives that depend on underlying financial assets like currencies or bonds.

Challenges in Financial Product Valuation

• The valuation process must consider counterparty risks post the 2008 crisis; this has led to increased regulatory scrutiny regarding CVA adjustments.

• Calculating CVA involves high-dimensional integration challenges typically difficult for classical computers but manageable through Monte Carlo methods.

Utilizing Monte Carlo Methods

• Monte Carlo methods are highlighted as essential computational algorithms that utilize random sampling for numerical estimations necessary for financial product evaluations.

• These methods face scaling issues related to accuracy due to fundamental limitations imposed by the central limit theorem affecting classical algorithms.

Quantum Computing and Monte Carlo Methods

Quantum Computing as a Solution

• Quantum computing presents an alternative to enhance the classical cost of scaling in Monte Carlo methods, particularly through quantum amplitude estimation.

• A classical algorithm requiring n steps for parameter estimation can be executed by a quantum algorithm in O(sqrtn) steps, demonstrating a near quadratic speedup over classical algorithms for specific problems.

Implications of Speedup

• For large values of n , such as billions, the difference between n and O(sqrtn) becomes significant, leading to substantial reductions in time and resource requirements for business applications.

Current Limitations

• Despite the potential advantages, current quantum devices are not suitable for near-term applications due to their inability to run deep circuits effectively.

Collaborative Efforts and Research Goals

• The partnership with Apata Computing aims to explore quantum techniques like quantum amplitude estimation and engineered liquid function techniques for credit value adjustment (CVA).

• The primary objective is to assess whether these techniques can provide advantages using noisy quantum devices while evaluating necessary hardware and computational resources.

Future Directions

• Positive experiences from this research have led to ongoing work on a paper that will detail results related to CVA estimates and analyze quantum speedup concerning noise and error parameters.