An interview with a quantitative researcher in finance
Based in Melbourne, Priya Mehra is a quantitative researcher applying behavioural finance to risk modelling in a global financial institution. This interview uncovers the practical impact of bias in technical environments, daily analytical routines, and advice for future quantitative analysts.
Describe your typical workday as a quantitative researcher in finance.
The assumption that numbers alone eliminate bias is flawed. Each morning begins by checking overnight market data, then running automated diagnostics on models built the day before. The biggest challenge is not technical, but interpretive—spotting where cognitive shortcuts, like confirmation bias, influence which variables are emphasised or ignored. Team meetings are rigorous: every assumption behind a model is challenged, sometimes line by line, to surface blind spots before models are used in live risk assessment.
What’s the most rewarding aspect, and the biggest challenge, in your role?
The highlight is constant exposure to new problems—no two days look the same. Cross-functional teams bring diverse views, and debate is encouraged. The toughest challenge is defending a model when market behaviour diverges from its predictions. It’s easy to double down on old assumptions, but that’s where risk creeps in. The team uses structured review sessions to force fresh analysis and avoid anchoring on past successes or failures.
How does cognitive bias affect risk modelling, and what steps do you take to control it?
Bias in risk modelling usually enters through model selection and data sourcing. Anchoring, availability, and overconfidence can all distort input weights or the selection of time periods for backtesting. The team counters this with a formal bias review: every model undergoes a step-by-step challenge process, with rotating leads to ensure no perspective dominates. Peer feedback is mandatory, and automated audit trails make it easier to catch recurring errors.
What advice would you give to someone starting out in quantitative finance?
For newcomers, the temptation is to trust the elegance of quantitative models. The best advice is to treat every output as a hypothesis, not a fact. Document every assumption, seek challenge from colleagues outside your immediate project, and routinely stress-test models against new data. Developing an instinct to question even your most cherished frameworks will keep both analysis and outcomes robust.
Can you share an example of bias affecting risk model outcomes, and how you addressed it?
In one project, we saw market volatility trigger a breakdown in an established model’s predictions. Upon review, it became clear that recency bias and selective data input played a role—recent events had been weighted too heavily, skewing results. The solution was to widen the input set and incorporate systematic bias checks into the development process. The lesson: robust risk modelling is as much about managing judgement as mastering mathematics.