Research & Publications
Research & Publications
Comprehensive quantitative finance research from the CU Quants community, exploring cutting-edge topics in algorithmic trading, risk management, and financial modeling
Embedded Files
Project Analyst Applications Now Open!
Exploring the Unexplored: Student-Driven Quantitative Finance Research
Exploring the Unexplored: Student-Driven Quantitative Finance Research
At CU Quants, research is our playground for discovery. As students passionate about quantitative finance, we tackle the niche ideas and overlooked concepts that often fall through the cracks of traditional academic and industry research. Our approach is fundamentally different—we're not constrained by commercial pressures or rigid academic frameworks, giving us the freedom to explore unconventional hypotheses and experimental methodologies.
Our research emerges from genuine curiosity and the desire to understand market phenomena that others might dismiss as too speculative, too complex, or simply not profitable enough to pursue. Whether it's analyzing obscure market anomalies, testing theoretical models in real-world scenarios, or combining disparate fields like satellite imagery with commodity pricing, we dive deep into questions that intrigue us.
Each project is a learning journey where we develop both technical skills and domain expertise. We're not afraid to fail, iterate, or pursue ideas that might seem unconventional. This student-driven approach allows us to investigate emerging trends before they hit mainstream research, explore interdisciplinary connections that professionals might overlook, and develop innovative solutions without the constraints of immediate commercial viability.
Our research reflects the questions we genuinely want answered, not just what's expected to be published. It's this curiosity-driven exploration that often leads to the most interesting discoveries and meaningful contributions to the quantitative finance community.
Our Research & Publications
Our Research & Publications
Efficient Quantum State Tomography via Burer-Monteiro Factorized Quasi-Newton Methods
Efficient Quantum State Tomography via Burer-Monteiro Factorized Quasi-Newton Methods
August 2025 | 1 Page | Magnus Miller | Community Spotlight
This work addresses a bottleneck in quantum state tomography (QST): scaling maximum likelihood estimation (MLE) to large quantum systems without the prohibitive cost of traditional constrained solvers. Standard approaches enforce quantum constraints directly, which leads to high memory use and slow convergence. Here the problem is reformulated via Burer–Monteiro factorization into an unconstrained, low-rank form that enables efficient quasi-Newton optimization with optimality guarantees. Two implementations are developed, a low-memory L-BFGS variant and a custom Gauss–Newton method, that maintain accuracy while drastically reducing computational overhead. This allows full tomography of 18 to 20 qubit systems in under 1.5 hours, matching or exceeding the performance of state-of-the-art methods such as CG-APG and MiFGD. The result is a shift from constraint-heavy, memory-bound formulations to scalable, structure-exploiting algorithms capable of handling quantum systems at a size most existing MLE methods cannot reach.
CAPO: Machine Learning-Driven Price Discovery for Competitive Market Making
CAPO: Machine Learning-Driven Price Discovery for Competitive Market Making
August 2025 | 18 Pages | Ryan Watts | Algorithmic Trading Team
CAPO tackles a neglected yet crucial problem in market making: how to price aggressively enough to remain competitive without hitting "toxic" levels that lead to instant adverse selection. Unlike static textbook models, CAPO combines machine learning with active exploration, using a dual Random Forest setup to distinguish "Safe" from "Toxic" prices while dynamically probing market boundaries under safety constraints. This approach captures real-time shifts in market behavior that passive models miss, allowing continuous adaptation in a microsecond-paced environment. Developed from a recognition of the gulf between academic theory and the harsh realities of electronic trading, CAPO addresses a technically narrow but strategically vital challenge most researchers avoid.
Google Sites
Report abuse