Section II: Potential Applications of Quantum Computing in the Securities Industry
In FINRA staff’s survey of quantum computing use cases, several firms indicated that they are examining how they can benefit from quantum computers when conducting various activities, including trade execution, portfolio management, risk assessments and fraud detection.34 Additional firms indicated to FINRA that they are assessing the potential risk quantum computers may pose to current encryption standards for securing critical data. Government bodies from around the world, including NIST, are attempting to address this potentially systemic issue related to data security by developing standards for post-quantum cryptography.
As market efforts related to quantum accelerate, the influx of attention and investment has led some market participants to worry about hype overtaking reality, with one group of researchers from the Massachusetts Institute of Technology (MIT) indicating that “quantum computing has a hype problem.”35 Some market participants have questioned the ability to develop commercialized applications for quantum computing in the near future. While the timeline for any practical changes resulting from quantum computing is a topic of ongoing debate, the theory behind how quantum computing works and its potential to solve large or complex computational problems is well documented.
Seizing on the potential associated with quantum computing, several financial institutions, including broker-dealers, have begun exploring how leveraging quantum for exponential improvements in computing performance could enhance business operations. Based on research and interviews with a range of market participants, the financial services industry has identified three potential areas involving computational challenges where quantum computing may have a significant impact: optimization systems, simulation systems and artificial intelligence.36
Optimization Systems
The ability of quantum computers to efficiently analyze and process numerous potential outcomes in real time may benefit optimization systems firms use. Market participants have indicated that computationally intense quantum models may process large sets of variables and allow for faster and more accurate optimizations, such as best determining valuations that can deliver competitive advantages; calculating more precise estimates of credit exposure; and better allocating capital across a range of corporate financial activities, among other uses.37
Financial institutions seeking to leverage quantum computing for enhancing optimization systems have focused on areas such as enhancements for trade execution, trade settlement and portfolio management. Market participants have indicated that quantum-based algorithms may be used to efficiently determine the optimal solution among various options38 because of the ability to survey multiple possible solutions simultaneously. Accordingly, quantum computers hold the potential to complete solutions for optimization problems in a fraction of the time it would take classical computers and firms are exploring its potential to more effectively navigate complex trading and investment environments involving large sets of variables.39 As outlined below, market participants believe that quantum computing’s capabilities for optimization may in the future streamline trade execution and settlement processes as well as enable investment managers to improve portfolio management.
- Trade Execution Optimization
Financial institutions are frequently tasked with determining how best to execute trades. Such efforts entail weighing a variety of factors, such as trade sizes, venues, timing and sequencing. These factors can impact the quality of trade execution and it is computationally challenging for even the most advanced classical computers to consider all the possible permutations given the number of potential variables. One report by a technology firm notes that “[t]o help put this type of optimization problem in perspective, consider that selecting the optimal order of execution for 5,000 trades has more than 4.2 x 1016,325 possibilities.”40 Some firms are assessing how quantum computing could offer a potential breakthrough for these trade execution challenges due to its ability to survey multiple possible solutions simultaneously, such as determining the optimal values for the sequencing, grouping and timing of trading activities.
- Trade Settlement Optimization
The trade settlement process involves securities being delivered by a seller against payment made by a purchaser, frequently facilitated by a clearinghouse, which can help to mitigate counterparty risk. Clearinghouses run trade settlement optimization analyses, which they use to determine the optimal routing trajectory for settling thousands of trades and matching up buyers and sellers at millisecond speeds, while also accounting for other legal, business, and operational factors.41 Given the different variables that need to be considered in the trade settlement process, determining an optimal solution may be computationally complex. Typically, there is greater complexity where there are more trades involved, which presents increased challenges for classical computers.42
In the view of some market participants, quantum computing may help optimize the trade settlement process by using quantum-based optimization algorithms to discover links for multi-party settlements, thereby making the process faster and more efficient. These market participants believe that a faster settlement process could potentially reduce cost risks related to replacement (i.e., the risk of loss from unrealized gains caused by delayed settlement); liquidity (i.e., market pressure that may result if firms fail to settle their positions); and credit (i.e., the possibility of loss where a party fails to meet its repayment obligations).44 Of note, the U.S. Securities and Exchange Commission (SEC) recently approved a rule change to shorten the settlement cycle to T+1,44 and reductions in settlement time may impact the benefits and risks associated with the potential future use of quantum computing to facilitate settlement.
- Investment Portfolio Optimization
Portfolio optimization involves determining the optimal mix of investment assets to achieve a desired objective, such as maximizing return and minimizing risk, over a given period of time. Investment portfolio optimization problems are frequently dependent on a number of variables and may at times result in calculations that take classical computers several days to conclude.45 As noted by some industry observers, the complexity associated with portfolio optimization can be attributed to a number of factors such as valuation adjustments (for credit, debit, funding, capital and margin); transaction costs; and regulatory and tax requirements; among others.46
Firms are researching whether quantum computing has the potential to improve the portfolio optimization process by offering the computational capacity to assess various scenarios involving a multitude of assets and related factors. As a result, some market participants believe that the use of quantum computers may at some point give firms a competitive advantage by allowing them to develop optimized portfolio options that are able to analyze more variables in a shorter timeframe and thereby better able to achieve the desired results. For example, quantum computers may potentially offer an enhanced ability to allocate weights to assets over a period of time (with the aim of maximizing returns), forecast asset returns, assess volatility and measure risk.47 In determining the optimal portfolio over time, quantum computers may seek to measure the return per unit of risk and do so while factoring in dynamic changes in the marketplace, measuring transaction costs and operating consistent with investment mandates.48
Simulation Systems
Quantum-based simulations could enhance firms’ abilities to understand and account for uncertainty related to market activity. Firms are examining ways to use quantum computers to run simulations of market-related activity that would otherwise be difficult or potentially impossible to capture with classical computers. In particular, financial institutions are exploring the use of quantum computers to assist with analyzing market activity related to risk assessments.
One such risk assessment method, the Monte Carlo simulation, which takes random samples of a number of variables to simulate probable outcomes, is a common technique for incorporating risk and uncertainty in financial models and evaluating potential risks.49 Firms may use a Monte Carlo simulation to determine the likelihood of possible outcomes for determining factors such as Value at Risk to help identify potential financial losses. The Monte Carlo simulation is also instrumental in pricing financial derivatives. For example, The Bank for International Settlements (BIS) has noted that, each year, more than $10 trillion worth of options and derivatives are exchanged globally, many priced using Monte Carlo methods.50 However, the Monte Carlo simulations typically use methods that are computationally intensive, especially in the face of an array of uncertainties, with some calculations taking classical computers several hours, or even days, to perform.51 Accordingly, firms are looking into ways to leverage quantum computers to more efficiently price complex derivatives—a task that can require a great deal of computing power, resources and time.52
Firms are engaging quantum computers to determine whether they offer the potential for a significant increase in processing power that could reduce the computation time for a typical Monte Carlo-based risk assessment from days or hours to near real-time.53 Some firms believe that a faster simulation process, provided by quantum computing, has the potential to provide benefits in circumstances, including regularly re-evaluating portfolios against risk factors such as liquidity and credit risk. Moreover, these firms are studying whether the greater processing power from quantum computers may also provide the capacity to consider additional market factors with the potential to improve the accuracy of risk assessments, enabling firms to better limit financial losses or enhance potential gains.54
In addition, some firms believe that quantum computing may assist with risk assessments related to anti-money laundering (AML) and know-your-customer (KYC) compliance systems. Quantum-based algorithms have the potential to improve AML and KYC programs by expanding the ability to analyze various elements of individual identity, transaction history, fund flows and relationships in real time. One relatively recent analysis of financial institutions suggested that even a one to two percent improvement in overall efficiency could save over $1.5 billion annually for the industry.55
Artificial Intelligence
As addressed in FINRA’s Artificial Intelligence (AI) in the Securities Industry report, the financial services industry, including broker-dealers, has allocated considerable resources to researching, developing and adopting artificial intelligence (AI) tools, including by leveraging computing power to analyze large data sets. Though quantum computing is still in a developmental phase, some market participants view it as a potential accelerant for AI because of its potential to enhance the ability to process and analyze large data sets.
The term “quantum AI” refers to a growing field in quantum computing that focuses on improving quantum-based computations and algorithms that are used to train models within various AI tools and applications. Some industry analysts have made the case that quantum computers may offer the ability to enlarge and enhance the type of datasets that AI algorithms and training models can use, resulting in a more efficient process and more accurate outputs, as well as helping AI models become more scalable.56 For example, quantum-based machine learning could potentially allow for the ability to process greater amounts of data while also analyzing that data at faster speeds, accelerating the pace at which machine learning models can learn.57 In addition, quantum-based natural language processing (QNLP) applies key quantum properties, such as superposition and entanglement, to allow deeper textual analysis and classification to train natural language processing models.58 Accordingly, some market participants are hopeful that quantum computing may be able to assist in the process of efficiently and accurately obtaining meaning and value from complex text and sentence structures to potentially assist with tasks, such as providing financial advice.59
However, as quantum computing may accelerate the potential beneficial impact of AI, it may similarly accelerate related risks. As noted in FINRA’s Artificial Intelligence (AI) in the Securities Industry report, model explainability and data bias are some of the key risks that need to be addressed when deploying AI-based tools, and these risks may be compounded by the complexity enabled through the use of quantum computing.60 As previously noted: “[f]irms that employ AI-based applications may benefit from reviewing and updating their model risk management frameworks to address the new and unique challenges AI models may pose.”61 This would be particularly true in the context of any quantum AI application.