What Can Quantum Computing Do in Finance?

Quantum computing is further along in finance than in almost any other field, and further from paying off than the headlines suggest. The theory is a decade deep, the largest banks run dedicated research teams, and there is still no quantum computer that beats an ordinary one at a real financial task.

This page is a map of that gap. What quantum finance actually is, the handful of algorithms it rests on, what the banks have really shown, and an honest answer to whether it works yet. Then the part almost nobody offers: three open, forkable projects from Qollab's Spring 2026 challenge that let you run a piece of it yourself.

The three projects

Three open-source quantum finance projects came out of the Spring 2026 challenge, each taking on a different classic problem. Quantum Systemic Oracle runs a portfolio-optimization circuit on IonQ and publishes the result on-chain as a risk score. Quantum Regime Radar uses quantum kernels to match live markets against historical stress regimes. Quantum Market Game turns the prisoner's dilemma into a two-qubit game where entanglement changes the outcome. All three are forkable, run on real hardware, and are candid about where quantum helps and where it does not.

Where quantum finance comes from

The idea is older than the hardware. In 1999, physicists Eisert, Wilkens, and Lewenstein showed that if you let players use quantum strategies, the prisoner's dilemma stops being a dilemma. The result comes with a caveat worth keeping: it holds only for a restricted set of strategies, and later work showed it does not survive once every quantum move is allowed. It was a hint, not a proof of advantage.

The finance-specific theory arrived in the 2010s. Ashley Montanaro proved a near-quadratic quantum speedup for Monte Carlo estimation, the workhorse behind pricing and risk. Rebentrost and colleagues turned that into the first quantum derivatives-pricing algorithm in 2018, and the 2019 Orus, Mugel, and Lizaso survey mapped the field: annealers for portfolios, arbitrage, and credit scoring, and amplitude estimation for pricing and risk.

Industry followed the theory. D-Wave, IBM, and IonQ put real machines within reach, and by 2020 a JPMorgan and IBM team had published a method to price options on a gate-based quantum computer. That short arc, from a game-theory curiosity to a bank paper, is the whole prehistory.

What quantum computers might do in finance

The use cases are well mapped, and each rests on one of a few quantum algorithms. The honest-status column is the part most write-ups leave out. Read it as promise, not product: most of these are quadratic speedups that only pay off on fault-tolerant machines far larger than today's, and on the machine-learning side, classical methods often keep pace.

Two patterns run through the table. The pricing and risk methods, built on amplitude estimation and quantum Monte Carlo, are real and provable, but the speedup is quadratic rather than exponential and needs error-corrected hardware. The learning methods, quantum kernels and QSVM, run on today's machines, yet on ordinary financial data a well-tuned classical model usually matches them, a pattern researchers call dequantization. The one exception that already matters is security, and it is a threat rather than a speedup.

Who is actually doing it

The clearest tell about quantum finance is that the bank doing the most is also the one publishing the doubts. JPMorgan's applied-research group, led by Marco Pistoia, co-wrote the 2020 option-pricing paper and keeps investing in the area. The same group also published numerical evidence against quantum-kernel advantage on classical data. A team that argues against its own hype is a good signal.

Others are running real experiments at small scale. IBM and HSBC co-authored a 2023 study using quantum kernels for fraud and credit classification, and in 2025 HSBC reported what it called the first known quantum-enabled algorithmic bond trading trial. IonQ and Fidelity's applied-technology center generated synthetic market data on trapped-ion hardware. In every case the authors call the results early and scale-limited.

Not everyone is leaning in. More than fifteen banks have active quantum programs, but the commitment varies, and Goldman Sachs, a co-author of the field's key resource-estimate paper, has reportedly scaled back its quantum team. The common thread is that all of this lives inside corporate research. None of it ships to a developer.

Does it work yet? An honest answer

No, not in the sense that matters. There is no demonstrated production quantum advantage in finance today. The speedups that do exist are quadratic, not exponential, and they need fault-tolerant machines far beyond current noisy hardware. The most-cited estimate, a 2021 threshold paper from Goldman Sachs and IBM authors, put pricing a real derivative at roughly 8,000 logical qubits and a circuit depth in the tens of millions, and called it out of reach of current systems. A 2024 method trimmed that estimate, but not to anything you can run this decade.

The machine-learning side has its own reality check. When researchers, including JPMorgan's own, test quantum kernels on ordinary financial data, a well-tuned classical model tends to catch up, and a 2024 review of quantum ML for finance concedes there is no provable exponential advantage for the near-term methods. Large value projections exist, such as BCG's estimate of up to $850 billion in economic value by 2040, but that is a forecast across industries, not revenue anyone is earning now.

One area is already real, and it is a threat rather than a speedup. A future quantum computer could break the encryption that protects transactions, so data harvested today could be decrypted later. That is why banks are starting to adopt the post-quantum encryption standards NIST finalized in 2024. One warning on names: a “quantum financial system” (QFS) and “Quantum AI” auto-trading platforms are scams and conspiracy theories, with no connection to any of the research above.

So where does that leave a developer who wants to touch this rather than read another projection? Not inside a bank lab. The three projects below are the accessible counterpoint, and together they map the whole on-ramp: a PhD whose dissertation is quantum finance, a bank data-governance veteran who had never run a quantum job, and a high schooler with a good mentor. None of them claims an advantage. Each is honest about its limits, which is exactly what makes them worth running.

A risk score smart contracts can read

Jamie Dominguez spent more than a decade in enterprise data governance at a global bank. The Quantum Systemic Oracle was his first QPU job and his first Qiskit run. It takes a portfolio-optimization circuit, runs it on IonQ, distills the result into a single systemic-risk index in basis points, and publishes that number on-chain through a Chainlink-shaped interface, so any smart contract can read it the way it reads a price feed.

The framing, in his own words, is quantum compute as an on-chain primitive: not a dashboard people look at, but a number other code is built on. He is candid that AI pair-programming is what closed the gap between his finance background and the unfamiliar quantum and blockchain stacks, and he treats that as the point, a working loop you learn from by running it.

Given that quantum advancements are showing promise for financial applications like portfolio optimization, it was a perfect use case to blend my interests.

Which kind of market is this?

A volatility model fitted in a calm market breaks in a crisis, so Alireza Khodaei built a tool that answers the prior question: which kind of market is this? Quantum Regime Radar scores live equity returns against five volatility regimes taken from real history, episodes like the 2017 melt-up or the run-up to the SVB collapse. Each regime is encoded as a quantum state, and a quantum kernel measures how strongly today's market overlaps each one. His doctoral research put exactly this GARCH estimation on quantum hardware, so the reference library rests on real backtesting rather than hype.

What sets the project apart is the honesty. Before claiming anything for the quantum side, the team built the classical case against themselves, a plain correlation study that topped out too weak to use, and shipped it in the project so you can see the baseline for yourself.

With any quantum algorithm, especially in finance, the first question you get is: everything is fine with classical, why even bother with quantum? We wanted to show, in one picture, that classical is not delivering a tangible advantage here.

Game theory, entangled

Aadarsh Venkat Ramanan, a rising high-school senior, found his way into quantum through Marco Pistoia, who heads JPMorgan's quantum research lab and had worked with his mother at the bank. His Quantum Market Game takes the classic prisoner's dilemma and runs it on a quantum computer: two traders are two qubits, each in a superposition of buy and sell, and an optional entangling gate links their choices. Turn entanglement on and the game settles into outcomes classical game theory cannot reach.

That idea has a serious lineage. The Eisert-Wilkens-Lewenstein scheme showed in 1999 that a quantum prisoner's dilemma opens up equilibria the classical game never had. Aadarsh keeps his version deliberately small, a two-qubit circuit you can read in one sitting, because it is built as a teaching object: superposition is the trader who has not decided, entanglement is the toggle that ties two fates together.

People should care because this is a baseline for future quantum use cases in finance, and it lets people learn the basic principles of quantum mechanics in an intuitive way, creating curiosity and further learning.