Demand by the buy-side for algorithms to trade forex is expected to continue to climb. That’s the collective view from bankers, vendors, industry analysts and developers contacted by FXALGONEWS. The algos are getting better and the focus on best execution has never been higher. That provides a recipe for more and more hedge funds, corporates and large asset managers to adopt algo-based trading strategies.
“Algos have become more sophisticated and much safer,” Philippe Buhannic, CEO of TradingScreen, told FXALGONEWS. “Also, algos have a tendency to be self-regulating, so they are constantly improved based on the effects of the last algo.”
The buy-side can look forward to a steady stream of new algos that can be tailored to their needs
Along the way, equities markets have offered some valuable lessons for algo developers and users as tools for the forex market have been refined.
One lesson has been the focus on the aggregation aspect of algos.
Thanks to the explosion of liquidity pools in equities after Reg NMS, it became important to track where execution took place. “Optimising through algos within market access is important,” Buhannic said. This is not an issue with futures, where liquidity is concentrated at one venue, but forex shares this trait with equities.
Buhannic states that 90% of his firm’s clients come from the buy-side and he can see changes in their behaviour.
“The most recent and biggest impact on the market has been that a number of large clients now realize that they can internally cross their FX volume,” Buhannic said. “As a result, clients have now reduced the volume they use to launch on the FX side, compared to what it used to be.”
Buhannic also said TradingScreen can see some of the buy-side now using internal crossing to generate a net position which they can expose to the market in any way they wish. This is mostly hedge funds but it can include larger asset managers.
“In some markets, we are now witnessing a growing interest in taking advantage of a specific shape of liquidity. Some of those interested are really sophisticated, and they have the benefit of not having to maintain a market. They really leverage that into a very selective participation within the market to generate an improved alpha.”
TradingScreen’s status as one of the largest TCA providers means it sees a lot of client reports. “I have seen some clients that have an abnormal capability of generating alpha. It’s more noticeable in markets with significant market structure issues.”
As an example, he cited Australia due to its position at the start of the global trading day. It is generally trading on old information and this can throw up opportunities for more sophisticated players.
Figure 1 – Internal or External Algorithmic Development
Buhannic said he sees more growth ahead for FX algos. He noted that while equity algos were responsible for about 60% of trading, the forex share was closer to 20-25%. “Algos are growing quite rapidly. On the equity side, it has stabilised quite a bit. I think it will stay that way. On the FX, side it will continue to grow.”
He believes growth will occur for two reasons. First, algos were getting better. “But more importantly, even if the algos were stale and didn’t improve, it will grow. And the reason for that is the best execution rule.”
Historically, he said, best execution did not really apply to forex or fixed income. But following on from events such as the Libor scandal, there is much more pressure on asset managers to ensure that they are getting good execution. That has led to some large lawsuits. “Nobody wants that in the middle of the market. The buy-side doesn’t want it, the pension funds don’t want it, and most of all, banks don’t want it. The only solution to avoid this is to have some kind of mechanism to measure best execution,” he said.
Still a role for humans
Amin Rajan, CEO of Create Research, agrees that algo trading will become more common, although he sees human trading still playing a big role. “Overall the trend is in the direction of greater adoption of algos, but it’s all in fits and starts,” he said.
Rajan said the buy-side can see both the merits and the downside to using algos. “However clever algos may be, algos cannot see the nuances of the markets as humans can,” he said.
For instance, Rajan said that algorithms work well when market conditions are calm, just as an airplane can fly safely on autopilot in good weather. But when markets are turbulent, it can be a different story.
As an example, he cited the Flash Crash, when the Dow Jones swooped almost 1,000 points only to recover within minutes. “People who had used stop-loss mechanisms – which very much rely on algorithms – ended up losing a huge amount of money,” he said.
Identifying signals in noisy markets can be an issue as well. “There are times when noise overwhelms the signal – to the extent that people end up making trades that are contrary to their best interests.”
Another important point for the foreign exchange market is that politics these days seem to be outweighing economic factors, Rajan said. In this respect, algos have a vulnerability since they are tools that are very much modelled on the past, and that can be less of a predictor for politics than for economic trends.
He said the best combination was using algos along with human insight. “They work in 90% of the cases they don’t work in 10% of the cases and those 10% end up being the more serious ones,” Rajan said. “Algos are not an all-weather tool.”
Development and Priorities
According to a recent FX Trading and Technology Survey 2014 which was published by Tibco, among those respondents who use algorithms, the percentage of those using in-house developed algorithms and bank-provided algorithms have both increased, especially on the buy-side. (See Figure 1) A point worth mentioning here is that it has become significantly cheaper to develop algos through the use of technology platforms such as TIBCO StreamBase® rather than via custom coding which has most likely contributed to the growth of in-house development.
Figure 3 – Top FX Capabilities to be Added/Improved by Buy-Side Respondents
Figure 2 – Biggest Data Challenge for Effective Algorithmic Trading in FX
There was also a deep concern amongst survey respondents for more accurate information—after all, an algorithm is only as good as the information fed into it. So “Reliability and Consistency of Sourced Data” has been viewed as the biggest challenge for algorithmic trading this year (See Figure 2). with “Algorithmic Order Execution” and “Auto Hedging” ranking top priority for improving FX trading capabilities over the coming year. (See Figure 3).
Increased FX algo activity
The evidence therefore all points to increased interest in using algos in the forex markets.Quantopian, an algorithmic trading platform that offers tools and data for people to develop, backtest and then live trade with their algos, has a unique perspective on the trend towards increased algo activity. “We see a lot of individuals and those individuals are looking to trade their own funds,” said John Fawcett, Quantopian’s founder.
Funded by Bessemer Venture Partners, Khosla Ventures, Spark Capital, Getco, and Wicklow Capital, Quantopian provides open source tools for developers. It has attracted thousands of users who are looking to create and refine algos but who lack the resources of developers at large institutions. Quantopian, having opened these resources up, are now in the process of developing a quant hedge fund that will leverage the top performing users in their community. Effectively that means they have the potential to inject a wave of new ideas into the collective algo pot.
At the moment, Quantopian is still focused on equity markets. But Fawcett said two of the top requests it gets from its user community are to add forex and to add currency futures. “The profile of users we’ve got are interested in FX or using it to explore strategies,” he said.
Whatever emerges from the open-source world of algos that Quantopian is facilitating could thus mirror the trajectory of algos developed by the sell-side in that they would begin from the universe of equity algos already in existence.
Competition among sell-side algo providers for buy-side clients is set to lead to yet more innovation. Currently, some algos offered by the sell-side are exploring adaptive approaches in the forex market, where they rapidly move in and out of different strategies to take account of changing liquidity characteristics. Some see the next wave becoming more interactive, where algos interact with the users and provide them with information that allows them to make adjustments in real time. All of this means that the buy-side can look forward to a steady stream of new algos that can be tailored to their needs or that they can interact with to potentially reduce impact cost. For intrepid buy-side players, these new algos will offer scope for gaining competitive advantages. But as with each wave of development, the market can expect new risks as well as issues from this next generation of algos get ironed out.