☰ Menu eFinancialCareers

Do traders need to know how to code?

Trading jobs coding

Time to learn Python?

You want to work as a trader. Does this mean you need to learn how to code? If you want to build an enduring career, the answer is almost certainly yes.

“If you’re simply a market maker or a discretionary trader, you won’t need to know how to code,” says Natalie Basiratpour, director at recruitment firm Selby Jennings. “But if you’re going into quantitative or electronic trading, then coding will definitely help.”

In other words, if you plan to spend your career working the phones, making markets on behalf of clients and deciding moment by moment how and when to place trades, coding isn’t a must. But if you want to work in electronic trading and to cede some of these decisions to machines, you’ll absolutely need to know that Python is more than a reptile. And with an increasing proportion of trades happen electronically, familiarity with Python and other coding languages looks pretty imperative.

Most junior traders now come ready-equipped with coding knowledge. “A lot of the guys who are coming through now have learned how to code at university and have electrical engineering or computer science degrees,” says Nathan Haynes, senior consultant in quantitative finance at recruitment firm GQR. “Coding is second nature to them.”

One electronic trader who’s spent time working in a senior role at Goldman Sachs, but asked to remain anonymous, says it’s a “generational thing”.  “Every computer science and maths student and every top university wants to be a trader, and they’ve all studied coding. Pretty soon, all the juniors in finance will be coding-literate.”

Cost-cutting is also likely to play a part in embedding coding at the heart of trading jobs. Historically, banks split quantitative trading teams into ‘quant traders’ and ‘quant developers’. Quant traders came up with trading ideas and quant developers translated those ideas into the code that would put them into action. When quant traders are proficient coders, this no longer makes sense. “Splitting the roles is very old school,” says the ex-Goldman trader. “I’ve worked in big banks that split the developer and quant roles and in smaller high frequency boutiques that don’t. It’s much more efficient when can code up your own trading model – it gives you an edge.”

Marco Bragazzi, finance director of high frequency trading firm Spire Europe, says the best HFT traders understand how their strategies are interacting with the servers at the exchange. The traders might not be expert coders themselves – Bragazzi thinks quant developers will still have this role – but they understand enough about coding to see how their trade ideas are being executed and how that can be improved.

While trading jobs become more development focused, the traditional role of the quant developer could become more about middle office maintenance. “You’re already getting developers who are doing more work behind the scenes – building analytics libraries and developing quantitative tools for the traders,” says Haynes.

Keen to maintain their employability, Basiratpour says some senior traders are going back to university and studying masters programs in computer science after eight or nine years in the industry: “There’s an awareness that people need to keep up to date with new methods.” High frequency traders need to know C++, C# or Python, says Basiratpour. Systematic traders need to know scripting languages like MATLAB and SQL.

The danger is that even recently acquired coding skills are superannuated by those of the latest batch of graduates. The ex-Goldman trader confessed that this is an issue. “I’ve worked in the industry for years and there are mechanical engineering graduates coming out of schools in Paris who know far more about some languages than I do.”

Ultimately however, he says coding is more about concepts than fine detail. “It’s a mindset. You need to wire your brain for problem solving using coding languages and tools. Once you’ve done that and learned one language properly, you can learn others and you’ll be a lot more employable.”

Comments (1)

Comments
  1. Python and Matlab are in the same category: interpreted languages which are considerably slower than any C dialect, but with lots of syntactic sugar and numerical libraries which make coding (not running the code) easier and quicker than with C++/C#. Sure, there are many ways to optimise Python and Matlab, but not everything can be optimised, and the stuff which can’t will run substantially slower than in C++/C#, which makes both languages an appalingly bad choice for high frequency trading. Those who claim otherwise, and who classify Python in the same category as C++/C#, are clueless. Also, SQL is not a scripting language in the same category as Matlab: SQL is a, well, the, database language, which is something completely different. Again, those who claim otherwise are utterly clueless. It is in fact quite worrying that such ignorant people recruit for such specialised roles: it means they are unable to understand the CVs they review.

The comment is under moderation. It will appear shortly.

React

Screen Name

Email

Consult our community guidelines here