Quant Reality Check

Inspiration

We began with a problem we had experienced ourselves: career titles are easy to find, but difficult to translate into real work and preparation decisions. Quantitative finance made this problem especially visible because trading, research, engineering, data and risk roles are often compressed into the same simplified image of “mathematics, coding and predicting markets.”

Our initial assumption was that students lacked quant-career information. Research changed that assumption. Official information was abundant, but fragmented across companies, job titles, recruitment stages and timelines. The real problem was interpretation and context.

Research and key insights

We collected 17 consented survey responses from a convenience sample consisting mainly of students and early-career explorers. Among the 16 respondents who received the revised information-gap question, day-to-day work and mathematics/programming requirements were the most frequently selected difficulties, with 11 selections each. Thirteen of those 16 respondents reported at least one concrete consequence of unclear information, including preparing low-priority material, making a poorly matched application, delaying an application or not knowing what to do next.

We also gathered two live interview perspectives—from a quantitative practitioner and a recruitment professional—and one asynchronous written student perspective. The practitioner explained that opportunities should be understood through company, role, recruitment stage and timeline, rather than treating “quant jobs” as a single category. The recruiter highlighted foundational implementation skills, while the student response showed how an early understanding of the field can over-focus on machine learning, automated trading and market prediction.

Finally, we reviewed 23 early-career roles across 8 selected firms, primarily using official job descriptions and recruitment pages.

We treated these findings cautiously. The survey was a small convenience sample, and an early form-validation problem allowed four respondents to select more than the intended three options. We therefore use the counts as directional evidence rather than population estimates. A sensitivity check using only compliant responses preserved the same three leading information needs.

What it does

Quant Reality Check helps first-time applicants move from a broad career label to specific questions they can investigate.

The experience combines:

An 11-question guided Reality Check covering career context, task preferences and current assumptions
Transparent workstyle signals across research iteration, real-time decision-making and systems building
A Reality Gap showing where a user’s expectations may be narrower than the reviewed evidence
A Preparation Gap connecting current preparation with recurring job-description requirements Exactly three contextual next actions to reduce information overload
A searchable library of 23 reviewed roles
Side-by-side comparison of roles across companies
Official source links, access dates and explicit “Not stated” labels
A Research Journey explaining how evidence changed the product
A downloadable, privacy-conscious results card

The workstyle signals describe current task preferences. They do not measure ability, personality or future career success. We deliberately avoided a black-box match percentage.

How we built it

We began with research rather than code. The team designed the survey and interviews, reviewed official job descriptions and organised 23 roles from 8 selected firms into a structured dataset. We recorded role families, responsibilities, skills, eligibility, location, recruitment status, sources and access dates while retaining each company’s original job title.

We then translated the evidence into product rules. Career-context questions determine which evidence and next actions are most relevant. Six scenarios generate transparent task-preference signals across research iteration, real-time decision-making and systems building. Two Reality Gap questions compare the user’s current expectations and preparation with the reviewed evidence. These signals suggest directions to explore; they do not measure ability or predict career success.

We used ChatGPT/Codex for AI-assisted coding, iteration and debugging. The prototype was implemented as a responsive React and JavaScript single-page application using HTML, CSS, SVG and browser Canvas features, then deployed on Netlify. The team remained responsible for the research design, source verification, product logic, implementation requirements, testing and final decisions. We manually tested the main result paths, role filters, comparisons, mobile layout and downloadable result card before deployment.

Challenges we ran into

One major challenge was turning inconsistent job descriptions into something comparable without erasing company-specific context. Similar titles can describe different work, while different titles can contain overlapping responsibilities. We addressed this by retaining official job titles, adding standardised role families and keeping every role connected to its original source and access date.

A second challenge was producing useful personalisation without claiming more than our evidence could support. Our small convenience sample and qualitative interviews cannot validate aptitude or predict career outcomes. We therefore removed “best fit” and black-box match percentages, used “direction worth exploring,” showed the reasoning behind each result and explicitly stated what the evidence could not prove.

We also encountered concrete data and implementation issues. An early survey setting failed to enforce the three-selection limit, allowing four respondents to exceed it. We retained the original responses, disclosed the problem and conducted a sensitivity check; the leading information needs remained unchanged. During final testing, we also found that two questionnaire screens displayed labels that did not match the values being stored. We corrected the mapping, enforced the selection limit and repeated smoke tests across the major user paths.

Accomplishments that we're proud of

We are proud that the project changed in response to evidence. It began as an information website, but our research showed that information already existed—the real gap was translation, comparison and context.

The final prototype brings together:

23 reviewed early-career roles across 8 selected firms
Official sources and access dates
An explainable 11-question Reality Check
Reality Gap and Preparation Gap analysis
Exactly three contextual next actions
Searchable role filters and side-by-side comparison
A transparent Research Journey
A responsive interface and downloadable result card

Most importantly, the product remains useful without pretending to be an authoritative career test. Users can trace recommendations back to their answers, reviewed job descriptions and clearly stated product rules.

What we learned

We learned that more information does not automatically lead to better decisions. Students need help understanding which information applies to their situation, what varies between firms and what must still be verified.

We also learned that responsible personalisation requires boundaries. “Not stated” is different from “not required,” a task preference is different from an ability score, and a small convenience sample should not be presented as evidence about an entire population.

AI-assisted development helped us prototype and iterate quickly, but it did not replace research judgment or quality assurance. Clear specifications, reliable evidence and human testing were still necessary. Finding and correcting the questionnaire mapping and validation problems reinforced that generated code must be independently checked against intended behaviour.

What's next for Quant Reality Check

Our next priority is to validate whether the product actually improves users’ understanding and decisions. We would run before-and-after usability studies with students and real applicants, measuring whether they can distinguish role responsibilities more accurately, identify appropriate preparation priorities and explain what they need to verify next.

We would also:

Add Australian and smaller quantitative firms to broaden the current large-firm sample Include more applicants with real recruitment and interview experience Establish a process for updating expired job listings and changing eligibility requirements Test whether different users receive understandable and appropriately cautious results Work with additional recruiters and practitioners to review role translations and preparation guidance

The employer contribution is a longer-term hypothesis rather than a proven outcome. If applicants understand roles more accurately, firms may receive fewer clearly misaligned applications and communicate opportunities more effectively. A future employer pilot could test this directly instead of assuming the benefit.

Our long-term goal is to develop the approach beyond quantitative finance into an evidence-linked career exploration framework for fields where job titles are visible but the reality of the work remains difficult for students to understand.