DLQF: The Dataism Laboratory for Quantitative Finance

Since its founding in 2024, the Dataism Lab for Quantitative Finance at Virginia Tech has become a cutting-edge interdisciplinary organization standing at the intersection of mathematics, computer science, and economics, dedicated to pushing the boundaries of opportunity in the field of quantitative finance for Virginia Tech students. DLQF aims to not only enhance students’ leverage in industry recruitment processes, but also strives to empower its members to conduct cutting-edge research & develop practical solutions for industry professionals in buy-side and middle-market banking. The lab is devoted to exploring the application of techniques at the forefront of machine learning, big data analytics, and econometrics to problems in quantitative finance in transformative and useful ways. The organization is composed of various teams conducting research and developing products day to day, with an admin team providing organizational structure and collective direction, headed by our director, Dr. Ali. We also work closely with external affiliates at other universities, such as Dr. Valavi at MIT. The lab has three primary sectors where teams operate. 

First and foremost is academic research, where teams conduct rigorous and pioneering research on a wide selection of topics within the quantitative finance sphere, ranging from the novel application of machine learning & statistical models to financial data problems, to the development of Python libraries for use in mathematical finance tasks. Our researchers are skilled in data preprocessing, statistics, programming, project management & technical writing tools like Overleaf. These projects ultimately focus on creating opportunities for members to publish in reputable academic journals, providing exposure in both academia & the industry, which are deeply intertwined in this field. One of the ongoing projects in this sector is the paper exploring the application of the CNN-Transformer architecture to financial datasets for prediction. The research team is novelly implementing a method used on time series data in other fields to stock data, combining CNN-based short-term pattern extraction with Transformer-based long-range modeling. This combination is meant to tackle one of the most prevalent issues in financial time series prediction: balancing short-term and long-term pattern extraction. The team is looking to publish their upcoming paper, including benchmark & trading simulation results, to finance and machine learning journals. 

Secondly, we have the client services sector, which focuses on developing quality, data-driven products tailor-made for the needs of external clients. These clients operate in various areas of finance, including investment banking, hedge funds, and private equity. The projects in this sector involve a greater deal of business skills such as creating pitch decks, document management, and weekly corporate reviews for healthy client communications & ensuring compliance. Technical skills in client-facing areas are also valuable in this sector, as creating client-ready products demands skills in UI/UX design & data visualization. The lab currently has a team working with Barret Capital Management, a hedge fund involved in quantitative trading, to investigate the effectiveness of ensemble modeling in forecasting for intraday momentum trading strategies. Open-Close-High-Low-Volume (OCHLV) data is frequently the only source used in traditional models, which may not adequately represent market dynamics. Thus, to improve prediction accuracy, this study incorporates non-traditional data sources such as interest rates, GDP projections, Twitter mood (calculated through

webscraping), and other macroeconomic indicators. The goal of the research is to increase signal robustness and adaptability in quickly shifting market situations by utilizing ensemble methodologies. Additionally, a team of DLQF members is working for Navagant, a middle-market bank that is rolling out two in‑house AI initiatives to speed up both deal sourcing and client‑facing deliverables. The first project automates prospect discovery: it reviews the firm’s historical engagements, identifies common attributes of past mandates, and then searches external market data to surface fresh companies with comparable profiles. A built-in scoring model ranks the most relevant decision‑makers and drafts outreach messages, allowing bankers to focus on high‑probability targets without having to build lists manually. The second project tackles presentation production. A vision‑based engine dissects reference slide templates, capturing every visual detail, and instantly rebuilds identical slides in HTML while inserting up‑to‑date financial figures. Early tests show the system can recreate complex layouts in seconds and greatly reduce the back‑and‑forth typically required for pitch‑book updates. Together, these tools are designed to shorten deal cycles, enhance analyst productivity, and reinforce Navagant’s reputation as a tech‑forward advisory firm without exposing sensitive information.

Lastly, there’s the algorithmic development sector, where teams design, implement, and refine machine learning algorithms for applications specific to pressing issues in quant finance. Some of the areas that lab members develop these algorithms for include trading strategies, risk management, portfolio optimization, and market analysis. Some of the lab’s ongoing projects in algorithmic development include: The application of CNN-Transformer architecture to financial time series forecasting & the application of deep learning techniques in financial settings. Members involved in this sector work with teams in both the academic & client side of the organization. All of these sectors and their endless potential for constructive experience and hands-on problem-solving are at the fingertips of DLQF members, with a solid backbone of experienced researchers and professors supporting and guiding their efforts into often uncharted territories. All projects in the lab fall under at least one of these categories, but are not limited to them and often overlap. These are the primary modes for members to gain skills, experience, and valuable resume items. 

An ideal DLQF member is a driven, curious professional with a passion for data, markets, and innovation. In the realm of technical skills, we look for proficiency in API implementation, data sourcing, JavaScript, Java, Python, time series analysis, and machine learning, with a preference for knowledge in linear algebra & mathematical statistics. The lab opens up applications every semester, so anyone interested in joining is welcome and encouraged to do so!