Rick Lane, CEO, Trading Technologies
If there is a sign that HFT has hit the mainstream, it's that the CEO of Trading Technologies, Rick Lane, is getting phone calls from his worried grandmother, asking if he's mixing with the wrong crowd.
After a brief stint at Google, Lane returned to Trading Technologies earlier this year to lead development of the firm's next generation trading platform. And though he is optimistic about the level to which low latency technologies have become accessible, Lane recognises that mass availability means regulatory oversight.
"This type of technology becoming more accessible and more available to the masses, you bring with that greater scrutiny," he said. "It is the same reason why my grandmother asks me if I read about 'the flash boys',
"That she is talking about HFT is shocking because two years ago that type of story would never have crossed her CNN channel."
For Lane, democratisation means that sophisticated technology is available to everyone, allowing everyone to operate on the same playing field. So a small floor trader who doesn't have an army of engineers on staff can compete in the low latency programming space with the large proprietary trading firms.
He entered the business almost a decade ago at a small prop firm as an engineer, later co-founding tech supplier, Tickit Trading Systems, from that experience.
"I worked with a bunch of traders who didn't know much about software engineering. Things would happen in the market in the middle of the day that they wanted to capitalise on and these were often very fleeting opportunities," he said. "Managing expectations on how long it actually takes to write code, make sure it is bug free and that we are all comfortable before we turn them on into production - it was just a tough environment."
Lane was part of a team that developed a platform focused on making the engineer obsolete, resulting in a kind of visual drag and drop method of constructing trading strategies. From there, a commercial platform was launched based on the technology - Algo Design Lab.
Years later, he still holds an idealistic technologist's view on the state of the market.
"Cost of acquisition is precipitously falling year after year, and it has allowed people to remain competitive," he said. "If you can be the first one to take an idea from just the idea conception to having something running, in a safe way, faster than anyone else, there is a real advantage."
Proliferation has also caused weaknesses in the market structure to be exposed related to "algos run amuck".
"People are learning their lessons, but the more automation you put into one of these systems and the more participants who are really just basing their trading strategies off of signals and letting the computers and robots do the work, the more opportunity there is for unforeseen emergent behaviour to occur," he said. "Exchanges have beefed up measures around kill switches…but we certainly have not seen the last of the major algo blow ups."
It's never been cheaper…
Huw Gronow, Head of Trading Europe and Asia, Principal Global Investors
Huw Gronow, head of trading for Europe and Asia at Principal Global Investors, said that regulators need to provide better ways of picking up on non-compliant trading behaviours in a microsecond context, and there is plenty of evidence that this responsibility is being taken seriously.
So after a period of easy set-up in an era when regulators promoted competition for new venues - leading to latency arbitrage opportunities - compliance requirements are expected to erect barriers to entry for those firms looking to capitalise on fragmented markets.
For its own automated strategies, Principal Global Investors has adopted algorithms via numerous brokers, which gives the firm a significant amount of control and autonomy over the heuristics of its execution.
But the overriding factor preventing firms from competing, said Gronow, is the "enormous" amount of spend required just to keep up in the low latency game, and there is also a relative burden on the regulator, exchange, broker and wider market. On the other hand, he also noted that "it is probably safe to say it has never been cheaper to post a limit order on a central limit order book".
He attributes the plunge in transaction costs to the march of technology that started some 30 years ago when Nasdaq opened its doors. Now, that same technological drive is moving past equities to FX, fixed income and commodities at quite a rapid pace.
"This is where HFT are now looking to expand into," he said. "HFT firms are continually looking for new opportunities."
High frequency execution
The shift into execution algorithms is one of the major trends that Stuart Farr, president of quantitative software provider Deltix, has been watching as sell side struggles because of diminished volumes. Largely staffed by engineers and mathematicians, Deltix has been exclusively focused on systematic quantitative trading since its inception and is therefore highly responsive to industry currents.
"High frequency has become less profitable…and over the last three years there has been a much greater focus on execution algorithms as opposed to signal generation algorithms," he said. "Everybody has cut their rates and there is only so low you can go,
"As a means of differentiation, (firms are giving) buy side a better execution experience through algos and more advanced (transaction cost analysis) tools."
For Farr however, the use of the term "democratisation" is somewhat disingenuous, and he prefers the term "commoditisation" to explain the way the industry is evolving.
"People have been able to mimic and achieve the same momentum or cross-over strategy that big firms are doing with technology that was expensive 10 years ago," he said. "High frequency trading is just very simple models that 10 to 15 years ago were difficult to implement that are now almost retail in terms of their availability."
Stuart Farr, President, Quantitative Software Provider, Deltix
Across the universe
The genie is definitely out of the bottle, said Bill Fenick, Interxion's strategy and market director for Financial Services, as HFT is moving into fixed income and commodities much more and becoming standard.
Interxion has data centres near major trading hubs throughout Europe. The firm's financial services customers are split almost evenly between prop trading firms with HFT strategies; big brokers and investment banks; and traditional hedge funds.
Fenick is part of the "cottage industry" that he sees springing up to support increasingly sophisticated trading technologies, which is spreading to microwave towers and lasers for firms that will make any level of investment to remain competitive on the level of speed.
"It's a huge population living off of this, (such as) data centres, Silicon Valley, chip makers, switch makers, market data providers, telecoms, OMSs, independent software vendors," he said. "It is a strong industry supporting this type of trading."
It's also an industry that might be entering a consolidation mode. When ICE acquired the New York Stock Exchange, one of its first moves was to hive off NYSE Technologies, which was built off Wombat's low latency feeds. That could be interpreted as being a sign that staying on the bleeding edge might not be worth the continuous investment.
"A leading exchange provider is saying this is not a market worth investing in for us," Fenick said. "I interpret it is as: this is a niche business again…there is not enough money there for the large companies."
Where developed markets may be saturated however, other jurisdictions might be stepping in as HFT democratisation spreads globally - the caveat being that it is ultimately an arbitrage play.
Interxion plays host to a number of Russian firms because the international order book sits at the London Stock Exchange, one of the data centre provider's major locus points. The IOB, Fenick added, is the largest market in Russian instruments outside of Moscow, with shares in Gazprom, Lukoil and Rosneft commonly the most traded stocks.
"Russia consolidated their two exchanges a few years ago and are trying to get more into this game, but you need somebody to arb against. If you are in Seoul Korea, who else is there?" Fenick explained, adding that Australian futures in Sydney is a market to keep an eye on.
"It will be interesting to see how (these and other) markets evolve," he said.
Just as HFT tools became cheaper to procure and integrate - such as complex event processing engines, time series data warehouses and mathematics libraries, for example - so too Deltix's Stuart Farr expected the same to happen to quantitative tools as well.
He is keeping an eye on the open source language programming movements in Python and R. "Free goes a long way to smoothing over decision processes," he said. "It forces us on the vendor side to be innovative and have a very compelling proposition…but (some firms) are going to struggle to maintain business models."
On the other hand, the cloud has enabled Deltix, among other firms such as Quantopian, QuantGo and QuantConnect, to get access to market data without physically shipping it around the world.
However, one of the consequences of greater access is that markets become saturated.
"People are going to have to become more imaginative in the use of quantitative ideas," he said. "Simple momentum and mean reversion strategies are not going to be sufficient."
Principal Global Investors' Huw Gronow also noted the advent of open source architecture in coding as well as artificial intelligence, machine learning and behavioural sentiment analysis as becoming more widespread. "I wouldn't be surprised if that is already comprising the next-stage genesis of electronic trading," he said.
So what lessons might be learned from the HFT experience?
Trading Technologies' Rick Lane said that a lot of the downsides from HFT were from the sheer speed aspect, and, if anything, some of the emergent big data storage and retrieval tools, such as NoSQL database technologies, are going to give advantages to the players focused on more than just low latency. Democratisation, therefore, is moving to making decisions based on data, not just speed.