Quant skills for discretionary traders

20170714 Retooling

Fashion is one of those things I think I might follow. In practice, the only times I end up being fashionable, is when the fashion world somehow meanders back to my taste. I used to wear a Casio G-Shock watch for many years, which I admit was hopelessly unfashionable for most of that time. Then all of a sudden, it became fashionable and I was in fashion. It’s like the old cliche of the broken clock, which will nevertheless be correct at least twice a day, when the hands of time happen to be at the same point as the clock’s stuck hands. I’ve been doing this job for exactly 12 years (well 12 years yesterday, on July 14th). Whilst, I’ve worked at various institutions, what I do has not really changed very much: basically looking at markets in particular currency markets from a more quantitative perspective. In most cases, it will be to create systematic trading strategies. For most of the earlier part of my career, I have to admit, this was not that fashionable (as we have already established, I’m never really one to follow fashion). In recent years, quant trading has suddenly come into vogue. You can’t pick up a financial publication without at least seeing a few stories about AI and machine learning and quants. One trend I’ve also noticed is for experienced discretionary traders to ask me the following question: how can we retool for the “quant age” in markets? This has particularly been the case in the past few months. Well, my first answer (talking my book here) is obviously to hire me as a consultant! However, aside from that what else can discretionary traders do to tool up for the quant age? I’ve put a few ideas below. I’m sure there’s loads of other things I could have suggested (if you have any ideas, which you think I should have added let me know!)


Experienced discretionary traders understand markets: don’t forget what you already know!

Experienced discretionary traders understand how the market works! By interacting in the market for many years, they will have picked up very particular knowledge about how markets trade and how they are impacted by events. Whilst, I still call myself a quant, I suspect this knowledge I have gained over the years is probably the most valuable part of what I know. After all, without this type of knowledge it makes it a lot more difficult to know where to start when it comes to building trading strategies. So if you’re discretionary traders and want to use more quantitative techniques, don’t forget what you already know. It’s not always the case that you can make discretionary strategies work in a systematic basis, but many can be. Also you will have a very good understanding of market liquidity, which is crucial.


Learn some Python: will speed up your analysis

Ok, I’m somewhat biased here. However, I think learning a bit Python can go a long way! Very often when backtesting trading strategies, you can start in Excel, but once the strategy starts to get a bit more complicated, Python is a good choice. It’s a lot easier to learn than stuff like C++, and already has lots of tools written for backtesting (like my open source libraries finmarketpy, findatapy and chartpy). I’m not suggesting you have to become an absolute expert in Python to number crunch markets (unless of course you want to). Even if you team up with other folks who will be implementing your trading strategies, it will at least be useful to know to understand what they’re doing. To get you on your way, there are video courses from websites like Coursera and Datacamp. I’ve also written some articles with hints and tips about online resources for Python, including this one.


Keep it simple: complexity is not always necessary

If you are just starting out on the route of quant strategies, it can be tempting to go to the most exciting technology, machine learning, AI etc. Just start simple, trying to model market behaviour you have already observed, and use that as a starting point for your backtesting. Simple techniques might work, without the need for something complicated. Again, just remember, it is about utilising your market knowledge to gain an edge, with technology as a tool (rather than the other way around). I hate the cliche of “low hanging fruit”, but this really is the case in trading. Market experience helps you find the orchards, which have the low hanging fruit!


Further formal education: which course to do (if any)?

One big question I’ve been asked by discretionary traders is whether it’s worth enrolling in quant based university courses? It really depends what you want to do, after all there is a cost to further formal education, both in terms of fees, but also time. Do you want to just understand the quant basics? Do you want to understand the details of software engineering and design whole systems? It of course possible to learn coding by teaching yourself and relying on online resources. However, I found it very useful learning coding and software design at university (I did a 4 year MSci degree in Joint Mathematics and Computer Science). In practice, I ended up unlearning many of the bad habits I had picked up coding there! My old university Imperial, for example has a 1 year Master’s in Computer Science for students who have not previously studied the subject formally, as a conversion course. In essence, it’s a crash course compressing many of the modules from their undergraduate Computer Science course. Other universities also do similar computer courses including Cambridge. It might also be the case that more data science related masters courses might be of relevant for what you do, such as NYU’s Data Science Masters. They will teach a mixture of coding and statistics, including work on machine learning. There are also financial engineering Masters courses, skewed more towards option pricing than to trading, which you could look at. Aside from Masters courses, there are also numerous boot camp style courses popping up, which typically last a shorter period of several months rather than a year.