How to write a home task report which is part of the interview process for a quant position in a trading firm

Question

I recently appeared in an interview for a quant research post in a trading company. As part of the interview, I was given a home-task to solve in a week. The inerviewer gave me a dataset consisting of high-frequency price, volume data for five stocks over a duration of four months and asked me to forecast volatilities of these stocks for the next month. I used a Bayesian multivariate stochastic volatility model for this purpose. My application was rejected one day after I submitted the results. The method that I used is an extension of the Philipov's Wishart autoregressive model and currently I am preparing a manuscript based on this method for publishing in a quantitative finance journal. I think it was the way I presented the method and the results that led to my rejection. Is there a format for writing these type of reports? Does anyone have an example of such report? Any help in this matter will be much appreciated.

(P.S. I am an experienced statistician with a PhD in statistical machine learning and 5 years work experience, but I am completely new to quantitative finance, so please forgive me if I sound too stupid in this post)

Answer 1

You sound like a good candidate for many roles in finance who probably has experience with some reasonable approaches to presentation. If your intuition is that you made some sub-optimal choices in the case of this assignment then you probably did, so try to get more specific about that and identify precisely where you believe you went wrong. On the other hand, you don't have to assume that your submission was really the deciding factor. What if they were also discussing the position with someone who had similar credentials, plus some big Bridge tournament wins, and the decision maker happens to believe that Bridge teaches everything about good decision making?

There's no agreement about appropriate presentation format. Always make it as clear and accessible as possible. You get more latitude on that point if you know your submission will be reviewed by someone who has academic credentials in your field. If it comes up again then you might ask the interviewer some questions about format and assess what would be ideal in the situation.

Answer 2

I've been heavily in the decision process of rejecting interview candidates at my firm. We have certainly come across PhD candidates who failed sheerly based on the poor quality of their presentation.

The application process is an art. It's a crapshoot. Even to your recruiters. Sometimes we already have a hypothesis against the candidate's favor by the time we offer him/her a chance to work on a 1 week take-home project. For example, a candidate sounds wonderful on paper (statistics PhD, list of programming proficiencies), but spams us with a list of 30 'papers' on his resume which are really just a mix of reports, undergraduate theses, technical white papers, meta-surveys of the field and conference summaries.

At this stage we already suspect he/she is the type to take credit for anything and don't really want to work with such a person, but nevertheless choose to administer a take-home test. Why? Because the marginal cost of administering a take-home test is nearly zero, but it gives us a chance to confirm the flaw we're rejecting him/her for. Perhaps his/her paper ends up with a working model for the problem, but say the tone of the conclusion suggests a much more important result than he/she has proven, it's already enough reason to reject the candidate because we have a vague confirmation of our earlier hypothesis.

(Don't hold it against your recruiters. False positives are much more costly to the firm than false negatives and most firms that have spent >2 years' time at this would have come to the same conclusion that they would much rather hire a less skilled candidate who is less likely to give them interpersonal drama in the office.)

It's the classic noise and signal problem. Basically, so much can happen behind the scenes (noise) that it's not possible for us to speculate based on your post. Your recruiters don't want to give any reasons for rejection that may be misconstrued or misrepresented as an unlawful, discriminatory action in court, so there's plenty of ambiguity (weak signal). The best you can do is to ask their HR rep nicely for some feedback.

As for examples of acceptable presentation skill, I imagine a MCM quality solution would have been sufficient for a 1 week take-home. (This is also why firms such as Jane Street and Two Sigma particularly respect candidates who have performed well at the MCM.)