On Friday, FXCM organized a full-day event inviting various speakers to discuss topics relating to algorithmic trading. After attending for most of it, here’s my write-up.
The entire day was put on FXCM’s expense account – so thank you FXCM for the invitation! FXCM has uploaded a video of the entire day (all 9 hours of it) on YouTube – view it here => Algo Summit – Live Broadcast
The venue was the Grange Hotel, a stone’s throw from the St. Paul’s Cathedral in central London. We were welcomed with fresh coffee and pastries at the 8am registration, a hot lunch buffett followed by more cakes and coffee in the afternoon. Top notch venue!
‘Algorithmic’ for retail traders
To most people, the term ‘algorithmic trading’ triggers visions of huge computers transacting at lightning speeds, fibre optic cables being re-routed into straight lines through mountain ranges, high frequency trading firms front-running stock prices, and the ‘Big Short’ movie. That type of algorithmic trading is out of the reach of retail traders as it requires huge levels of investment into IT infrastructure and dozens of PhD’s to find and develop those tiny edges that are then traded thousands and thousands of times per day.
In the context of the FXCM Summit, the term means ‘systematic trading’ – building automated trading strategies that can be written with code such as Python, back-tested on platforms such as MultiCharts or TradeStation, and trades being executed on a retail platform such as FXCM. There is a big community of retail systematic traders, and numerous books have been published targeting the retail market. I follow a fair bunch of these guys on my Twitter feed so checking out my ‘Following’ list can be a good starting point.
In this sense, systematic trading is an alternative to manually looking for trading setups and entering trades manually by pushing buy and sell buttons. Many retail traders complain about the difficulty in following their trading rules and being disciplined. Adopting a systematic trading approach can partially address these issues though it does require other skills to be learned such as coding and proper back-testing.
Yves Hilpisch – a popular author and speaker in this area (Python for Finance) who discussed the increasing role that Artificial Intelligence (AI) is likely to have in the trading space going forward. After getting to some slide which he purely intended to use as an example of some back-testing results he had obtained for one of his trading strategies, he was inundated with a huge amounts of questions, which I think reflected that attendees were keen to get into the nuts and bolts and trying to get answers to challenges that they were encountering in their own trading – to me, it also showed that many of the attendees had not progressed significantly far down the road of systematic trading and were still trying to get ideas of the ‘secret sauce’ from traders such as Hilpisch. Great speaker – extremely knowledgeable.
Stephane Ifrah – after giving an insight into his start-up ventures in the crypto currency sector, Ifrah gave his view of what quantitative finance (aka systematic trading) is all about and how to go about building a systematic strategy. In his view, it is very important to start with an idea based on observing the markets, articulating and defining the rules for the idea and then back-testing – as opposed to starting with large amounts of raw data and letting computers (or AI) look for patterns in it. Critical success factors for him are data quality, filtering systems for reducing trading frequency whilst maintaining profits and finally appropriate money management that takes into account both volatility as well as strategy edge (cf. Kelly Criterion).
As Ifrah was speaking, it helped me to rethink my approach with my existing automated strategy – how to improve it – and also ideas for different types of strategies.
Rob Carver – the keynote speaker – similar to Hilpisch, Carver patiently answered a large number of questions – his view was similar to Ifrah, in that he preferred designing strategies on ideas rather than finding patterns from raw data using machine learning. Carver delivered an excellent presentation with some very great one-liners along the way. After warning of the dangers of over-fitting automated strategies and outlining three different types of ‘overfitting’ – some of which traders can do so unknowingly, a great deal of the presentation revolved around testing various parameters for building a trend-following strategy on fake/random data and then running the parameters over real market data. Presentation slides. After listening to his talk, I will certainly go out and buy some of his books.
Arthur Sepp – I have to admit most of his presentation went well over me head. Sepp went through a large number of charts in which he explained correlations between trend-following strategies in various market conditions and then used some trading strategy examples with FX futures contracts. He was able to do in the same comfort as I would talk about my weekend. He did appear incredibly smart, and his mannerism, physical build and impeccable self-presentation certainly made me feel that I would never want to be on the other side of his trade. He also runs a blog artursepp.com
FXCM Workshop – An FXCM staff member outlined the various options available for FXCM clients looking to get into systematic trading, many at zero or very low cost. This includes an API to enable trades to pull live and historical data on all markets offered by FXCM, the ability to write strategies and to deploy tested strategies and have them trade live on the FXCM platform. This was more detailed and practical than anything discussed in the other presentations. I’ll be following up with FXCM staff to see whether I could get my coded trend-following strategy (all done in TradeStation using EasyLanguage) onto the FXCM platform. TradeStation, although maintaining live FX price feeds on their platform, no longer offers trading in spot FX, and secondly, the FX data sometimes has inaccurate price spikes which can distort back-testing as well as calculations for determining trade setups.
Unfortunately I had to leave around 3pm thus missed the Q&A session with Rob Carver as well as the presentation by Andreas Clenow on how to go become an industry professional.
All in all, a great day out and a good exposure to new knowledge and ideas.