A hedge fund in the US has invented a new system that could make it much harder for traders to spoof high-frequency trading and destabilise markets iStock

A hedge fund in New York has come up with an innovative way to high-frequency trading (HFT) fraud – a system which combines neural networks, computer science and an incredibly accurate atomic clock that can sync orders to within a few billionths of a second.

HFT is a programme trading platform which uses supercomputers transact a larger number of orders at very fast speeds, while complex algorithms analyse multiple markets and execute orders based on current market conditions. Traders that execute orders faster tend to be more profitable than traders with slower execution speeds.

Unfortunately, the technology is controversial as some dealers carry out dubious trading activities that affect the stability and integrity of equity markets by suddenly pushing prices up or down, and many governments and firms now believe high-frequency trading should be subject to more regulations or even banned.

There are several controversial trading practices which enable rogue traders to manipulate markets, such as flash trading – traders pay a fee to see incoming orders for buying or selling securities about 30 milliseconds earlier than the rest of the market, using the information to make a profit; or insider trading – where they have access to non-public information about companies.

There is also spoofing, whereby traders collude to make huge futures bids on commodities – such as barrels of oil – but they have no intention of actually buying them, and instead are seeking to make other participants believe in an illusion of demand and profit by cancelling the order when the price changes.

How the anti-HFT system works

Renaissance Technologies, an investment management firm that specialises in systematic trading performed using only mathematical and statistical quantitative models, was founded by James Simons, an award-winning mathematician and a former Cold War codebreaker.

The firm has filed a patent application with the US Patent and Trademark Office for an anti-HFT prevention system. This features a neural network of computer servers, complex algorithms and an atomic clock that uses an electronic transition frequency in the electromagnetic spectrum of atoms to help it keep time.

Atomic clocks are considered to be the most accurate timekeeping devices in the world. They work by being precisely calibrated to vibrations of irradiated atoms of a soft silvery-gold alkali metal called Caesium.

According to the patent, the solution works by having a trading server that divides a large transaction order into multiple smaller transaction orders and gives each of the orders a transaction execution time. The trading server then sends a financial trade instruction based on each smaller transaction order to a set of relevant secondary servers.

Each of the servers is fitted with an atomic clock, and as soon all the clocks reach the set transaction execution time, all the servers then submit their much smaller transaction orders to the respective financial exchanges simultaneously. This helps to prevent spoofing, which is one of the hardest crimes to catch.

Preventing spoofing would stop the threat of rogue traders

In November 2015, trader Michael Coscia became the first person to be prosecuted in the US over spoofing. He was convicted of commodities fraud and spoofing for illegally earning $1.4m (£1.1m) in less than three months in 2011. Coscia entered more large orders than anyone else in the world, in multiple markets ranging from corn and soybeans to gold that he didn't intend to execute.

Then there is the ongoing case of London-based trader Navinder Singh Sarao, who is fighting extradition to the US to stand trial for allegedly helping to cause the 2010 flash crash through spoofing. This saw Wall Street shares tumble within minutes as tens of billions of dollars were wiped off share values, while Sarao made a profit of $875,000.

If the orders are made collectively by a group of traders, it is harder to detect that spoofing is going on while the supply, demand and price of the commodities rise or fall, so if the solution were to work, it would be able to prevent an artificial inflation or deflation in prices.

It would also help prevent situations like the unfortunate 2009 case of a City of London trader who got drunk and placed a trade for 7 million barrels of crude oil, which caused an artificial spike and not only cost him his job but also a £72,000 fine from the UK Financial Services Authority.