One of our goals at Quantopian is to provide quants with the skills and tools they need to be as smart and successful as possible. One of the things we’re doing to help the community is bringing in a variety of experts who can share information and insights to make Quantopian as awesome as possible. We’re super excited to introduce our newest advisor, Wes McKinney.
Many of you may be already familiar with Wes. He’s a bit of a rock star in the Python community. His recent book, “Python for Data Analysis,” discusses the nuts and bolts of manipulating, processing, cleaning and crunching data in Python. It’s a fascinating read, with content for new users just getting starting, as well as Python masters looking for detailed tips and tricks. He’s also the co-founder of Lambda Foundry, which provides high-productivity solutions for data science and helps drive forward pandas, data analysis and visualization.
Wes began his career as a hedge-fund analyst, just one week after the August 2007 quant equity crisis. But he worked through it and learned a lot about research and the tools of the trade. Along the way, he became interested in building better data tools for Python.
I spoke with Wes recently to talk about his career, pandas, Python and his thoughts about being part of the Quantopian community.
You started your career as an analyst at a hedge fund, can you talk a bit about that experience?
It was a really fun place to be, and very interesting – especially since I started in August of ’07, a week after the quant crisis started. It was a very collegial environment and not siloed at all. I worked with people from many different backgrounds, including economics, CS, and pure math, which is how I ended up discovering Python. I enjoyed the work, but ultimately left because I was interested in finding new challenges, which led me to grad school.
You might be best known as the author of pandas. What motivated you to start such a huge project?
Well, it wasn’t a huge project when I started! I was using R and dealing with a lot of gnarly time series and cross-sectional data. As a result, I started encountering all sorts of data alignment headaches. I realized I needed a customized solution, and I felt that Python was the right language for it. By building table-like structures with automatic data alignment on top of NumPy, I ended up with something that a lot of my colleagues wanted to use.
And what about pandas today?
Pandas has seen a lot of uptake in the financial sector, especially quantitative asset management. A lot of companies are looking to get away from legacy Java applications, and as pandas has matured, it’s made the data management side of things much simpler. It’s hard to gauge exactly how much it’s been adopted in the industry, but it’s safe to say there’s quite a bit of it out there.
These days, you’re a pretty prolific author on Stack Overflow. What keeps you so active there?
It’s about answering people’s questions. I really like capturing the bugs that surface there. The people on Stack Overflow are pretty savvy, and they’ve all invested a lot of effort in their various areas of interest. Looking at the examples and solutions that come up there helps me find areas where I can improve pandas’s functionality or API. And it’s really gaining traction – a year ago there was only one or two question per day, and now it is far, far more.
And speaking of being an author, congratulations, you recently published “Python for Data Analysis” with O’Reilly Media. Who should read it, and do you have plans for another book in the future?
I wanted to write a book that would work for two audiences. The first are people who are new to Python, or at least are new to using Python for data wrangling. There really wasn’t a good resource on how to get started doing this, and the documentation pages that did exist weren’t especially satisfying. A lot of people were asking questions along the lines of, “I use R; where do I get started with NumPy and pandas?” So my book tackles that.
On the other hand, I use this stuff every day, and in the process, I’ve learned quite a bit. Some of the content goes pretty deep, so it should be interesting for expert users as well – particularly people who use IPython, as it explores a lot of cool features there.
To be honest though, I’m not in a hurry to write another book, but I might. It was stressful at points, but most of the time it was fun. It was definitely fun finishing! In general, though, I prefer hacking to writing.
What advice would you give someone interested in becoming a quant?
For aspiring quants, I suggest obtaining an advanced degree or working at a hedge fund to learn practical skills and applications of theoretical skills. While some of the entry-level work – such as data cleaning and debugging production processes– may be unpleasant at times, I believe it provides the fundamentals for success.
What advice would you give someone just starting out in this field?
First, get your head around the practical skills and understand how the theoretical skills can be applied in real situations. For most people that means getting an advanced degree or spending time at a hedge fund. When you’re just starting out, it’s really sink or swim. A lot of people interested in becoming quants are hot-shot math types, and sometimes they can be disappointed to learn that a fair amount of the practical work isn’t that sexy. You know, basic stats, linear regressions, data cleaning and keeping systems up and running. It’s probably a lot simpler than what they had been doing in school, but at some level, it’s the foundation. The value-add comes in deciding which combination of tools to use in order to build a solid model; it isn’t necessarily about building the fanciest model possible.
What was the appeal of getting involved with Quantopian?
There are a number of things that make Quantopian an interesting idea to me. First is the whole philosophy, you know, the idea that quantitative trading algorithms can be open and shared and that lots of people can learn from them. The whole community thing is also very appealing. You asked me about Stack Overflow, there’s something great about being able to engage with people and help them think through challenges. And, of course, to work with Fawce and the Quantopian team.
What will you be doing here at Quantopian?
You tell me! I expect to be working on a couple of fronts. First, I’ll be helping with the codebase – reviewing code and analyzing it for performance bottlenecks. I’ll also be speaking at a few of the meetups, discussing Python, financial analysis and sharing a bit of my own investment philosophies. If people are in Boston or New York, they should come and check it out – even if I’m not the speaker.
Look for Wes to be appearing here and there in the Quantopian ecosystem. You can learn more about him by visiting his Quant Pythonista blog. If you have any questions for Wes, please feel free to reach out to me at firstname.lastname@example.org , and I’ll be sure to pass them along!