How to Get a Job at PDT Partners

PDT Partners is a systematic quantitative hedge fund based in New York City. The firm was originally founded as Process Driven Trading within Morgan Stanley in the early 1990s by Peter Muller, one of the most respected figures in quantitative finance. PDT operated as a proprietary trading group within Morgan Stanley for nearly two decades before spinning out as an independent firm in 2012, following the Volcker Rule's restrictions on proprietary trading by banks.

PDT manages several billion dollars in assets using purely systematic, model-driven strategies. The firm develops statistical models that identify predictive signals in financial data and uses these models to trade across equities, futures, and other instruments. Unlike discretionary hedge funds where portfolio managers make subjective judgments, PDT's trading is entirely automated — human researchers build and refine the models, but execution is handled by algorithms.

The firm is known for its exceptionally long and successful track record. PDT was consistently one of the most profitable groups within Morgan Stanley, and it has continued to perform well as an independent fund. The team is small — approximately 100-150 employees — with a high concentration of PhDs and world-class researchers. PDT's research culture is often compared to an academic department, with the key difference that ideas are tested against real financial markets with real capital at stake.

PDT's culture is best described as academic, rigorous, and deeply collaborative. The firm operates more like a research lab than a typical hedge fund — researchers are given significant freedom to explore ideas, the atmosphere is intellectually stimulating, and the pace is more measured than at high-frequency firms. PDT values deep thinking and careful research over speed and aggressive risk-taking.

The firm's roots in Morgan Stanley's proprietary trading group have shaped its identity. PDT has a long institutional memory — many researchers have been with the firm (or its predecessor group) for over a decade, creating a wealth of accumulated knowledge about what works in quantitative finance. New hires benefit from this deep expertise through mentorship and knowledge transfer. The firm actively encourages collaboration between researchers, and there is a genuine culture of intellectual generosity.

Work-life balance at PDT is generally good by industry standards. The systematic nature of the firm's trading means there are no market-hour-driven emergencies or adrenaline-fueled trading floors. Researchers work on longer time horizons than traders at market-making firms, which creates a more sustainable pace. The firm values quality of research over quantity of output, and employees are encouraged to take the time needed to do rigorous work. For candidates who want to do deep quantitative research in finance without the frenetic pace of a trading floor, PDT offers an unusually attractive environment.

PDT seeks candidates with exceptional research ability and strong quantitative foundations. The ideal candidate can develop novel statistical models, work rigorously with data, and bring deep expertise in machine learning, statistics, or signal processing to bear on financial prediction problems. The majority of PDT's researchers have PhDs in quantitative fields — statistics, computer science, physics, electrical engineering, or applied mathematics.

Beyond technical skills, PDT values intellectual curiosity and research taste — the ability to ask good questions, identify promising research directions, and distinguish signal from noise in both data and ideas. The firm wants people who are intrinsically motivated by the challenge of prediction, who read papers for fun, and who have a track record of independent intellectual achievement (whether through academic publications, competition results, or innovative projects).

PDT also looks for strong programming skills in Python and/or R. Research at PDT involves working with large datasets, implementing complex models, and building robust data pipelines. Candidates need to be comfortable writing production-quality code, not just notebook-level prototypes. Experience with large-scale data processing, SQL databases, and cloud computing infrastructure is a plus. Finally, PDT values collaborative researchers who can communicate clearly, engage constructively with colleagues' ideas, and contribute to the firm's knowledge-sharing culture.

PDT's interview process typically consists of 4 to 5 rounds over 4 to 8 weeks. The process is heavily focused on research ability, statistical thinking, and technical depth. PDT takes its time with hiring decisions — the firm would rather miss a good candidate than make a bad hire, which means the evaluation is thorough and deliberate.

The general structure is:

• Recruiter/initial screen (1 round): A brief conversation about your background, research interests, and motivation for PDT. This is partly informational — helping you understand the role — and partly evaluative, assessing whether your profile aligns with PDT's needs.
• Technical phone screens (1-2 rounds): 45-60 minute conversations with researchers covering statistics, machine learning, probability, and programming. These go deep on specific topics rather than testing broad but shallow knowledge. You may be asked to solve problems in real time or discuss past research in detail.
• On-site interviews (2-3 rounds): A half-day or full-day at PDT's New York office. Multiple rounds with different researchers covering statistical modeling, machine learning, research methodology, and coding. You may be asked to present past research or work through a case study involving data analysis and model building.
• Final round: A conversation with senior leadership or the CTO focused on research direction, career goals, and cultural fit.

PDT's interviews are notable for their emphasis on depth and rigor. Interviewers will push until they find the edge of your knowledge, and they value honest acknowledgment of uncertainty over confident bluffing. The firm is looking for people who think carefully, question assumptions, and approach problems with genuine intellectual honesty.

Each stage of PDT's interview process targets specific competencies:

Statistical Modeling and Inference: PDT's core business is statistical prediction, so expect rigorous questions on regression, time series analysis, hypothesis testing, and estimation theory. You might be asked to discuss the assumptions behind linear regression, explain how to handle non-stationarity in financial data, or derive the properties of a particular estimator. Interviewers value depth — they want to see that you truly understand the methods you use, not just how to call them from a library.

Machine Learning: Expect questions on both classical ML and modern deep learning. You should be able to discuss the bias-variance tradeoff, explain regularization techniques, compare different model architectures for time series prediction, and reason about when different approaches are appropriate. PDT values practical wisdom — understanding not just what works in textbooks but what works in practice with noisy, non-stationary financial data.

Research Methodology: PDT will test your ability to design and evaluate research. You might be given a hypothesis and asked how you would test it, or presented with experimental results and asked to identify potential issues (look-ahead bias, overfitting, confounders). The firm wants researchers who can self-correct — who anticipate problems before they corrupt results.

Programming and Data Skills: Coding interviews at PDT focus on practical data science skills rather than competitive programming. Expect to work with data in Python — cleaning datasets, implementing models, computing statistics, and producing visualizations. Clean, readable, well-organized code is valued. You may also be asked about software engineering practices: testing, version control, and writing maintainable research code.

Research Presentation: You may be asked to present a past research project — either from academia, industry, or personal work. Prepare a clear narrative: what was the problem, what did you try, what worked, what didn't, and what you learned. PDT evaluates both the quality of the work and the clarity of your communication. Be prepared for detailed follow-up questions that probe the robustness of your methodology.

PDT's interviews test statistical knowledge at a depth that catches many candidates off guard. You need to go beyond familiarity with methods to genuine understanding of why they work, when they fail, and how to adapt them to non-standard situations.

Study "All of Statistics" by Larry Wasserman for a rigorous overview, and supplement with "Time Series Analysis" by Hamilton for the time series methods central to financial research. Make sure you can derive key results — the OLS estimator, the properties of maximum likelihood, the bias-variance decomposition — from first principles. Practice explaining these derivations clearly, as if teaching them.

Focus especially on topics relevant to financial data: non-stationarity, regime changes, heavy-tailed distributions, and multiple testing corrections. Financial data violates many textbook assumptions, and PDT wants researchers who are aware of these challenges and know how to handle them. Work through problems that require you to identify and address violations of standard assumptions.

PDT values practical ML experience over theoretical knowledge alone. The best preparation involves working on real prediction problems with messy data — ideally in a domain where signal-to-noise ratios are low, similar to finance.

Build projects that demonstrate your ability to go from raw data to deployed model: data cleaning, feature engineering, model selection, hyperparameter tuning, evaluation, and interpretation. Kaggle competitions (especially time series prediction or tabular data challenges) are good practice, but go beyond just maximizing a leaderboard score — document your methodology, discuss what you learned about the data, and show awareness of practical deployment considerations.

Study the ML methods most relevant to quantitative finance: gradient boosting (XGBoost, LightGBM), neural networks for sequence data (LSTMs, Transformers), Bayesian optimization, and ensemble methods. Understand the tradeoffs between different approaches and when each is appropriate. PDT values researchers who can choose the right tool for the problem, not just apply their favorite algorithm to everything.

Review PDT compensation data and book a coaching session for personalized research interview preparation.

You will likely be asked to present past research during PDT's on-site interviews. Having a well-prepared, compelling research narrative significantly strengthens your candidacy.

Choose a project where you can clearly articulate: (1) What problem you were solving and why it matters, (2) Your methodology and why you chose it over alternatives, (3) Key results and insights, (4) Limitations and what you would do differently. The project doesn't need to be in finance — academic research, Kaggle work, or personal projects all work — but it should demonstrate rigorous statistical thinking and careful methodology.

Prepare for deep follow-up questions. PDT interviewers will probe your understanding: What assumptions did your model make? How would results change if those assumptions were violated? How did you validate your results? What's the biggest weakness of your approach? Practice presenting to friends or colleagues who can ask challenging questions and push back on your claims. The ability to defend your work while remaining intellectually honest about its limitations is exactly what PDT evaluates.

For the highest-quality mock research presentations, Quant Blueprint's coaching program connects you with mentors who have worked at elite systematic funds like PDT. Our team of 10 quant traders and researchers will challenge your methodology the way PDT interviewers do, helping you refine both your statistical reasoning and your presentation delivery until you can confidently defend your research under intense scrutiny.