If anyone thinks speed is dead, says Graeme Burnett, a high-performance computing expert, then best they look for a career outside of the financial markets.
FX is becoming a world of stream computing, FPGA, GPUs, and an evolving computer architecture that is constantly advancing new capabilities to deal with complexity.
“Strategies in FX have been threshold-based…now we have the ability potentially to do real-time analytics. Rather than having a test facility to get those analytics computed, we can bring them closer to the strategy,” says Burnett.
The traditional way that FX was traded is in the process of changing fundamentally…
That means being able to react to signals much quicker than before in an era when electronic trading is taking over.
What’s interesting about the foreign exchange story, is how quickly that’s happening compared to other asset classes, says Donal Byrne, CEO of Corvil, an IT data analytics firm that works with platform and market data providers among its FX clients.
“(FX markets’) rate of adoption of technology is a lot faster, like two or three times faster, in terms of time, than has traditionally been the case,” he says. “That bodes well for opportunity, but it also has its own challenges.”
Byrne predicts a compete dis-aggregation and rebuild of the FX market: “The traditional way that FX was traded is in the process of changing fundamentally, and I do think that’s pretty exciting because I would anticipate we are going to have a much more efficient FX market place.”
When it comes to machine learning and artificial intelligence capabilities, however, a high degree of adaptation should be expected.
“Our view is that everyone is going to start using faster, smarter machines to allow the machines to trade autonomously, and it’s a combination of machine learning, big data, and really super-fast connectivity…companies that are able to leverage those three technologies are going to gain competitive advantage,” he says.
…there’s so much information and signal feeding what a currency’s (relative value) should be…
One of the firms blazing a trail on this front is CargoMetrics, a 7+ years-old technology company that uses quantitative methods to build investment programs from proprietary data. It is fully machine-driven, 100% systematic, non-discretionary, and is combined with human oversight to ensure the systems are operating within parameters and as designed.
The edge, says Scott Borgerson, CEO of CargoMetrics, comes from deep expertise in maritime data, and a methodology that is unique, proprietary, and patented. The firm performs trillions of calculations daily across several markets: commodities, fixed income, equity futures and FX.
In a nutshell, CargoMetrics is about a “technology-first approach, and the pursuit of using big data, cloud computing and satellites to have better information about monitoring and mapping global trade”, Borgerson says. “And then building better quantitative signals out of those data in a systematic approach.”
It is largely commodities-focused, he adds, but initial forays into FX have opened Borgerson’s eyes to the complexity of the market.
“Of all the markets we trade, it’s one of the harder ones…because there’s so much information and signal feeding what a currency’s (relative value) should be, besides the fundamentals,” he says.
It’s about having a holistic understanding of what’s happening in the market
Axia Extends Market Footprint in GCC RegionGo to article >>
The advance of artificial intelligence in FX is an exciting development for traditional traders, and using the right tools is a big challenge, says Victor Lebreton, director of Quant Hedge, a trading firm that deals in liquid FX pairs among other asset classes.
“We are systematic traders…more short-term, medium-term frequency guys, trading from several minutes to several hours…so capturing the right price is good but catching the right movement in the market is even better,” explains Lebreton. “(FX) is now coming into a global forecasting environment…and if there is some real information coming up in the market, then your models can refocus the complex financial market environment.”
Increasingly, firms are making investments into artificial intelligence on both the macroeconomic and microstructure sides – in option pricing for example.
“They want to recompute in real-time forecast of pricing. It is very heavy computing and demanding because artificial intelligence models need to recompute all the information behind to make sure it has the right optimization,” says Lebreton. “We are only at the beginning from our side, we are not doing it on a large scale, but I think that is the most exciting thing in the markets one year or two years from now.”
…artificial intelligence models need to recompute all the information behind to make sure it has the right optimization…
Holistic market views
Though artificial intelligence is a dominant theme, so too are big data and cloud computing in terms of how the trading environment is changing and creating opportunities, says Bartt Kellermann, founder and host of Battle of the Quants, the premiere event in the quantitative finance industry.
This year’s London “Battle” will tackle the debate between traditional quants, who rely on price and volume, and the new believers in unstructured data ⎼ social media and news analytics ⎼ as part of generating alpha.
Some expert opinions point out that although diverse data sets are interesting, they are also expensive. So, unless you are trading on a lot of money, and making a lot of money, the alpha may not be justified from a profit perspective, Kellermann explained.
It is true that in any type of model, typically price has been the most important factor, for trend followers, for example, says Saeed Amen, consulting quant and founder of Cuemacro. Still, there are other sources to augment price signals in determining different perspectives of markets.
Brexit, for example, showed just how much politics can become a driver of the GBP.
“It’s about having a holistic understanding of what’s happening in the market,” says Amen. One of the most exciting developments, he adds, is the way the financial industry is trying to understand how social media is impacting markets.
But it’s not the case that only looking at social media signals will win the day either. It’s about seeing unstructured data as an additional tool that can be used to understand the markets better, Amen notes.
You have to have more intelligent machines that can watch over and safeguard…
Corvil’s Byrne adds a cautionary note: “I am excited about machine learning, big data, cloud technology, virtualisation, all those things, but part of our task is to make sure that those things work and work correctly…You have to have more intelligent machines that can watch over and safeguard the intent of what those algorithms were supposed to do.”
Scott Borgerson will be delivering the keynote address at Battle of the Quants in London on 16 November, 2016. Battle of the Quants looks at quantitative finance developments across all asset classes.