Some of my friends and family wonder what on earth I do day-to-day. I figured a blog post on this ‘important activity’ makes sense.
I tend to rise early, between 5am and 5:30am on most days. My day starts with a glorious cup of black coffee for Emi (my partner) and me using the infamous AeroPress coffee maker.
I then read for 30 minutes and then either go for a 5k run (Tuesdays, Thursdays, and Saturdays) or just get ready and have breakfast.
With the morning routine done, and a healthy, nutritious breakfast eaten, I start my work day.
The pre-Office routine
For about an hour – hour and a half, I look at my code / programmes that were either completed the previous day or are still in progress.
If there are any minor ‘bugs’ that I didn’t identify earlier, I deal with these then and there before letting it continue to run. I then make my way to the office on campus.
On days we run, I take the bus to campus. Otherwise, we walk – it’s about 50 minutes door to door. During the walk, Emi and I talk about our weekly plans (Mondays), reviews (Wednesdays), and reflections (Fridays).
She joins me for the walk until about halfway through to campus. For the remainder of the walk, I either listen to podcasts, audiobooks, or some music courtesy of Amazon Prime.
When I take the bus, I typically read one or two articles on Towards Data Science or check out what people are posting on LinkedIn with hashtags for #datascience, #finance, or #python.
At the office
Once I’m at the office, I start by checking my email and respond to any important ones then and there. After that, I don’t look at emails until after lunch.
Work during the day
Work during the day varies quite a lot. For instance, this week I’ve mainly been:
- Compiling regression results into LaTeX tables.
- Plotting and compiling graphs for some of the results.
- Writing up the ’empirical results’ section of my paper.
Last week predominantly involved:
- Running Fama MacBeth regressions on my dataset.
- Testing ‘linear factor models’ on my ‘tradeable factor’.
- Compiling summary statistics for Table 1 of my paper.
Generally speaking, the weekly tasks / goals tend to vary. And I don’t really maintain a rigid structure of the workflow since things change quite dynamically.
While getting data, majority of my time goes on collecting and cleaning (“pre-processing”) the data to a stage where I can use it to drive meaningful insights.
From the 2nd year of my PhD onwards, I’ve been packaging my data cleaning programmes so I can reuse the code and avoid having to do a lot of the manual work repetitively.
As with most people who work with empirical data, a lot of time is spent on googling around for solutions.
This ranges from ways to process the data in a specific way (e.g. sorting firms into specific ‘bins’, creating a “3D plot” of ‘double sorted portfolio returns’), as well as statistical methodologies (e.g. how to accurately measure test statistics in a specific regression).
StackOverflow is a God-send, and I do try and spend some of my time giving back to the community by answering some questions. I must admit though, for the most part, I’m lurking around to find solutions to my problems!
Reading. Lots of reading.
Depending on what stage I’m at with my paper, I could be reading anywhere between 3-5 papers a day (during the initial ideation stage), to somewhere closer to 3-5 papers a week (during the data collection stage and beyond).
I keep an eye out for interesting papers on SSRN, and of course read papers published on the most reputable journals in my field including The Journal of Finance (JF), Journal of Financial Economics (JFE), Review of Financial Studies (RFS), and the like.
For the most part, I avoid textbooks since they’re often out of date by their very nature. That being said, I do have a handful of ‘go to’ textbooks for statistical methodologies.
These are particularly important because I don’t work with off-the-shelf statistical packages like STATA. I prefer having full, end-to-end control of my paper’s methodology, which is why I predominantly work on Python.
Continuous Learning Time
I do tend to allocate roughly 20% of my time on learning something new. Recently, I’ve been focusing on getting a hang of LaTeX.
I was almost always apprehensive about it since I found Microsoft Word just so much easier. But I finally caved in because the quality of the font on LaTeX is unparalleled.
Plus, I’m finding that it helps me write with better focus.
Prior to that, my 20% learning time involved exploring Spark for Big Data on Python, using Docker to create ‘containers’ for my code, exploring new methodologies in NLP, etc.
This 20% learning time is usually reserved for Thursdays or Fridays, depending on the workflow for that week.
My go to resource for on-demand learning is Udemy. In fact, it was Udemy courses (and a whole host of projects) that helped me learn Python from scratch. I’m eternally grateful to Jose Portilla for his incredible courses, and have talked about 2 other people I’m really grateful to in this post.
Seminars and ‘Brown Bags’
During term time, the Finance Group at WBS hosts weekly seminars, inviting academics from a whole host of Ivy League Universities including UC Berkley, Harvard, Stanford, etc.
‘Brown Bags’ also tend to take place weekly, sometimes fortnightly, wherein PhD researchers and academics can present their working papers (i.e. work in progress research).
I particularly value both of these, because firstly it’s free knowledge (who could say no?!). Secondly, it keeps me in the loop of the most current and up to date research that’s going on.
Comments during seminars and brown bags can be ruthless, and utterly brutal. But this is a really good thing. Because you get meaningful feedback for your work – not sugar coated nonsense.
Teaching, marking, and ‘office hours’
I typically teach one term every academic year. During that term, I’ll teach undergraduate and / or postgraduate students for between 4 and 6 hours per week.
I’m usually given up to 3 hours to prepare for the tutorials, plus some complicated way of determining the number of ‘office hours’ (support outside of class hours) I give students. Please don’t get me started on this ‘calculation’.
I intentionally hide my email address from the Learning Management System (LMS), but students somehow manage to find it anyway.
When they email me with content related questions, I happily reply – sometimes the questions are particularly interesting and intriguing. I ignore all emails relating to admin; things like “I can’t see my exam schedule on my timetable” (me neither, mate.)
In the summer (technically, UK does have summer), I have some marking duties, usually involving exam scripts. I’m fortunate that most of the exams I mark are quantitative ones, so determining if an answer is ‘right’ is pretty straight forward.
It always pains me to see an exam script where a student has done all the work correctly, only to have gone and scratched the whole thing and asked me to ‘please omit’. Truly painful.
I meet with my supervisor approximately once a week, sometimes fortnightly. We go over the work in progress, hiccups, updates, and plans for next steps.
These usually last for about an hour. I’ve found meetings to be most productive when I’ve got something to show him. Mainly because he goes out of his way to give me incredibly detailed feedback, line by line, for my work.
That’s as far as formal meetings go.
Thanks to the fact that I’m surrounded by people who are a lot smarter than me, I often have short but really productive ‘chats’ with fellow PhD researchers in our open plan office.
These typically last for between 10 and 20 minutes, and end up saving me at least an hour’s worth of googling around.
Back at home
I leave the office between 4pm and 5pm (the last bus is at 4:45pm, I kid you not).
Once back home, I spend some time with Emi before heading up to my study / home office and continuing with my work. I’ve setup Remote PC so I can access my office computer from home.
By about 6:15pm I start wrapping up, starting any code that I want to have running overnight, and reflecting on the tasks I worked on that day.
We have dinner for 6:30pm, and chill with some Netflix for about an hour. By 8pm I’m back in the study while Emi gets ready for bed. I spend some time reflecting on what went well that day, and what could’ve gone better.
I make a few notes in my Bullet Journal before closing shop and calling it a night.
And that’s it!
By no means does all of this happen every day. Seminars and brown bags take place once a week each, maybe fortnightly. Meetings with my supervisor are also at about the same frequency.
I suppose the thing that’s consistent pretty much every single day is coding. And this is the part I most enjoy of my PhD in all honesty. That, and finding answers to interesting questions of course!
You can read a bit about my current research here.