New Year flash currency crash seen as ‘perversion’

The flash crash originating with the Japanese yen that shook currency markets on 2019’s first day of trading was likely caused by algorithms.

The yen surged against the US and Australian dollars between the Wednesday close in New York and Thursday’s open in Tokyo.

Known as the witching hour among traders, and exacerbated by the public holiday in Japan, its a period of extreme market illiquidity – a market in which it is difficult to sell assets and in this case caused a lack of buyers – and the market became dominated by machine trading.

The lack of trading activity in other nearby markets conspired to provide a gap where buy and sell orders can go in one direction, moving the markets sharply until there is sufficient liquidity to make the trade. When other computers pick up on this action, a flurry of computer-driven trading can occur.

Machine-led trading now dominates the US stock market and some analysts think algorithms have led to recent selloffs, with the impact becoming more visible throughout 2018. According to Statista, RoboAdvisors manage nearly US$1 billion worth of assets.

Other analysts and traders have been speculating that the

Apple profit warning, due to dwindling iPhones sales in China, and which preceded the rally by just an hour, was the culprit.

Other factors that could have been catalysts are fears of a slowing China, the US government shutdown now in its second week, and investors buying up yen, long seen as a ‘‘safe haven’’ currency.

Kiwibank senior trader, Mike Shirley, who was on shift when it happened, said it was largely due to algorithms.

‘‘There was a lot going on at the time but essentially this was all just noise in the background. Really it stemmed from algorithm trading and Japan was on holiday.

‘‘Physical people are [normally] putting numbers in to the market, but the machine spat the dummy and we saw massive swings in price [of the kiwi] as high as $67.20 and as low as $65.80 in just a few moments.’’

The kiwi hit a high of nearly 73 against the yen on Thursday morning before falling sharply down to 70, but was up again Friday to 72.50.

And Nick Tuffley, chief economist at ASB agreed, saying ‘‘[machine trading] is a lot faster but not very well designed to cope with the illiquidity’’.

‘‘For the NZ dollar, essentially we got very briefly punched down quite hard, particularly against the yen.

‘‘Against the US dollar we are certainly firmer overnight. We didn’t really move that much against the euro. So some extreme moves from a very short period of time and most of its settled back,’’ he said.

Machines are trading on behalf of investment banks and have changed the way price is discovered through artificial intelligence, high frequency trading and headline ‘‘word triggers’’.

Big moves can trigger them to try to correct. And sometimes they can oversteer, which like on Thursday, perverts the market.

Shirley, who was watching in real time said it wasn’t the reaction you would expect to see.

‘‘My initial reaction was to think that someone keyed in a wrong number,’’ said Shirley, before realising that was unlikely.

He said computers are unable to pick up on contextual nuances.

‘‘They have pre-defined structure whereas a physical person can say ‘that’s wrong’.’’

‘‘There was a lot going on at the time but essentially this was all just noise in the background.’’

Mike Shirley