Quantum Computing: Reshaping Quant Finance Beyond the Flash Crash

Crisis on the Edge: The Flash Crash and the Dimensional Disaster of Classical Computing

On May 6, 2010, the Dow Jones Industrial Average plummeted nearly 1,000 points in under 20 minutes, with the steepest drop of 600 points occurring within just five minutes. Nearly $1 trillion in market value evaporated before the index sharply recovered.

U.S. law enforcement investigations identified the culprit as Navinder Singh Sarao, a futures trader.

He used a self-developed program to place massive fake sell orders on U.S. stock index futures, creating artificial selling pressure to manipulate prices. This induced other investors to follow suit and sell off their holdings. He then rapidly canceled his orders. After the index crashed, he bought back in at lower prices, selling off once prices recovered to profit from the volatility. This high-frequency trading strategy, known as spoofing, involved placing massive sell orders in a short period to manipulate the index, causing the shortest financial panic in U.S. stock market history—the infamous Flash Crash.

This leads to a critical question: If a top-tier institution manages a diversified portfolio of hundreds of assets, and the market suddenly encounters a more sophisticated version of Sarao—who might use advanced algorithms to hide manipulation traces or even leverage AI to generate dynamic fake order flows, triggering market volatility far more severe than in 2010—how should institutions with massive capital make their decisions? Should they sell off in panic or hold steady?

When the market experiences an abnormal crash, avoiding becoming a victim of panic selling requires precisely assessing the potential downside risk of hundreds of assets in subsequent market evolution and the correlation risk transmission paths between them within an extremely short timeframe. For institutions, the core competitive advantage of investing has never been merely the ability to make money, but rather the ability to make money while keeping risks under control. The cost of failed risk control can be devastating.

Therefore, in extreme market conditions similar to the "Flash Crash," institutions must quickly answer a series of questions:

• What is the overall risk exposure of the current portfolio?
• What is the maximum possible drawdown for each asset?
• Which assets have strong correlations that could trigger a domino effect if one drops?
• Should high-risk assets be sold off directly, or hedged through derivatives?
• What proportion should be sold to control risk without triggering a secondary stampede?

Behind each question lies the complex computation of hundreds or thousands of variables. To obtain answers within the "golden minutes" of market panic, extreme computing power is required. The problem is that traditional classical computing has already hit a dead end known as the "dimensional disaster" in such high-dimensional risk assessment tasks. This is precisely the opportunity for quantum computing to step onto the financial stage.

Computing Power as a Dimensional Strike: The Technical Foundation of Quantum Computing

Quantum computing is a computing technology based on the principles of quantum mechanics. It utilizes the superposition and entanglement states of qubits to rapidly process large-scale data and solve complex problems that traditional computers struggle with. The core of quantum computing is the qubit (quantum bit). While a classical bit exists in a state of either 0 or 1, a qubit can exist in a superposition of both states.

Quantum Superposition: The Key to Exponential Parallelism

This means that N qubits in a superposition state store information regarding an exponential number ($2^N$) of binary configurations, collectively forming a quantum state. When an operation is performed on any single qubit among the N qubits, the entire quantum state is manipulated, indicating massive parallelism.

However, current quantum computing is still in the exploratory phase.

Hybrid Computing: Engineering Compromises in the NISQ Era

First, the coherence time of qubits is limited, making it impossible to run long sequences of fully quantum logic. If the entire machine learning workflow were handed over to a quantum computer, the quantum state would quickly collapse due to noise, rendering the calculation invalid.

Currently, a mainstream approach to mitigating quantum noise is the adoption of Noisy Intermediate-Scale Quantum (NISQ) devices. These devices do not attempt to correct noise but instead strive to operate within strict limitations imposed by the noise. IBM’s Heron processor is an example of a NISQ device. Furthermore, quantum computers currently cannot directly handle data cleaning, model deployment, and other engineering tasks. These tasks rely on the software ecosystems of classical computers (such as Python), which quantum computers cannot yet replace. Therefore, current quantum computing typically requires interaction and integration with classical computers, representing a hybrid approach.

Taking Root: From the Laboratory to Trillion-Dollar Financial Practice

HSBC and IBM: The World’s First Quantum-Empowered Algorithmic Trading System

On September 25 this year, HSBC and IBM collaborated to release the world’s first known quantum-empowered algorithmic trading demonstration system. This system utilizes cloud-based quantum processors to handle real-world data from the European corporate bond market, valued at $12 trillion. The results showed a 34% improvement in the accuracy of predicting trade execution probabilities.

Quantum computing has long been hailed as the next frontier, but it was often dismissed as hype without practical application scenarios. Today, it is delivering tangible value in the financial sector. This development may not only reshape bond trading but also transform the entire quantitative finance architecture, echoing the transformative role large computers played in stock exchanges during the 1960s and 1970s.

Back then, IBM’s bulky machines revolutionized stock exchanges by automating data processing and supporting early algorithmic strategies. Today, IBM’s Heron quantum processors are beginning to tackle problems that traditional computers struggle to conquer. History may not repeat itself, but it is certainly rhyming.

Watching IBM stand at the forefront of technological change once again, we cannot help but sigh. As a former tech giant, IBM’s presence has been relatively quiet in recent years, often giving the impression of a company in decline. However, this century-old enterprise, when you don’t see it, may simply be quietly preparing a major breakthrough. It seems to possess an indestructible gene.

Comprehensive Penetration into Portfolio Optimization and Risk Management

Beyond bond trading, quantum technology is extending into various sub-fields of quantitative finance, promising disruptive progress in areas constrained by traditional technologies.

For instance, in portfolio optimization, the Quantum Approximate Optimization Algorithm (QAOA) can handle the explosive growth of asset allocation combinations, incorporating constraints such as risk tolerance and correlations faster than traditional solvers. A 2024 paper on quantum machine learning highlighted applications in this field, pointing out the potential of this algorithm to enhance returns in multi-asset portfolios.

Additionally, risk management will benefit significantly. Calculating Value at Risk (VaR) and predicting tail risks traditionally requires running tens of millions of Monte Carlo simulations, which are time-consuming and computationally expensive. The Bank for International Settlements’ report on quantum opportunities in finance mentions that quantum amplitude estimation accelerates these simulations by quadratically reducing the required sample size.

For derivatives pricing (options, swaps, and exotic bonds), quantum technology can revolutionize Black-Scholes Extension Models, enabling real-time pricing of complex path-dependent instruments and potentially reducing calculation time from hours to minutes.

Computing Power Democratization: The QaaS Model Opens an Inclusive Era

Given the myriad applications of quantum computing, can ordinary people afford to use it?

Breaking Cost Barriers: From Millions of Dollars to Pay-As-You-Go

Frankly speaking, quantum computing is expensive. The cost of quantum computers has long been the core bottleneck hindering their industrialization. The high costs stem from three major areas: hardware, maintenance, and R&D. In terms of hardware, the dilution refrigerator required for superconducting quantum computing can cost over $800,000 per unit. Coupled with the precision manufacturing of quantum chips and the铺设 of microwave control links, the hardware cost of a thousand-qubit system can exceed tens of millions of dollars. In terms of maintenance, quantum computers require operation at temperatures close to absolute zero, with annual electricity and equipment maintenance costs reaching millions of dollars, further raising the barrier to entry.

However, with the rise of the QaaS (Quantum-as-a-Service) model, the low-cost billing method of $0.1 per quantum hour is frequently mentioned. Whether this price can truly achieve widespread adoption has become one of the most watched topics in the quantum computing industry.

The core logic of the quantum-as-a-service model is to dilate the cost of quantum computers for individual users through computing power sharing and scaled operations. In the traditional model, users bear the costs of整机 procurement, maintenance, and upgrades alone. The QaaS platform connects quantum computers to the cloud, allowing multiple users to call computing power on demand, significantly improving equipment utilization rates.

Currently, some enterprises have introduced billing models based on the number of quantum gates or coherence time, with the cost of some simple calculation tasks dropping to a few dollars per hour. As the number of connected users increases and technology iterates, factors such as improved yield rates in quantum chip manufacturing and optimized cooling system costs may drive billing prices even lower, providing a technical foundation for the goal of $0.1 per quantum hour.

Figure Caption: