The automated trading environment, and the technology that underpins it, is coming under serious scrutiny, and with good reason.
First of all and algo ran amok, almost bankrupting Knight Capital, and then last week the Tokyo Stock Exchange (TSE) and Bolsas y Mercados Espanoles (BME) both suffered technical glitches that halted trading on them.
The TSE said that a problem with the router in its Tdex+ derivatives trading system was the cause of the 93 minute outage on its derivatives platform.
For the Japanese exchange this is becoming an alarmingly regular problem. In February 2012 a breakdown in its data distribution system halted trading on 241 securities and in December 2011 it suffered a blockage in network cables that slowed down execution.
At the BME it is still less clear what the problem was, although recently the bourse has been implementing new technology infrastructure that has made its trading system ten times faster and plans to introduce collocation facilities.
Glitches inevitable
The truth is though that glitches such as these are inevitable because it isn’t economically justifiable for exchanges to ensure otherwise.
To make their systems completely resilient requires exchanges to change them as infrequently as possible, which causes massive redundancy in these systems. Meanwhile today’s markets are moving and evolving so quickly that the exchanges are under pressure to constantly upgrade their systems in order to make them faster or introduce new products or order types.
Essentially, the amount that it costs an exchange to have 100% availability over a period of one year compared to 99.9% availability is significantly more compared to what they might lose from an hour or two of trading outage during this time.
The idea that exchanges will lose business to rivals during such an outage may not even be that accurate.
Implications for exchanges
Steve Grob, director at Fidessa, said that in the equity markets when similar glitches have occurred that this certainly hasn’t been the case.
“In Europe if the LSE goes down or the NYSE cash markets go down, people don’t trade BATS and Chi-X and Turquoise, they just don’t trade,” he said.
According to Grob the reason for this is twofold. Firstly, people are reluctant to trade on other platforms because of the psychological discomfort of trading on a stock that’s primarily listed on an exchange that isn’t producing prices. Secondly, the market maker’s systems are predicated from the start with the price on the primary exchange and use that as a basis for making that price.
Despite this, Keith Ducker, chief investment officer at TORA, argued that the TSE outage could have serious ramifications for the bourse.
Ducker said that not only will this latest technology malfunction dent confidence in market, but it could also affect the proposed merger between the TSE and the Osaka Securities Exchange.
“At the very least the TSE’s valuation will likely take a hit. Also, one of the primary selling points of the transaction was costs saving in technology. It appears that perhaps they have been too cost focused in this area, let alone cutting costs further within a combined entity,” he said.
Regulator’s role
Both Grob and Ducker claimed that the regulators have a part to play in these outages.
“If someone needs to step up and take the blame for this it’s the regulators, not the exchanges,” declared Grob.
“Regulators have created an environment that has encouraged fragmentation and competition, which in turn has created a breeding ground for HFT. So as a net by-product of trying to make markets fairer they’ve made them more complicated,” he added. It is these resulting complications that have lead to exchange technology failing.
Ducker said that regulators need to step in and hold exchanges to a higher standard of performance, introducing financial penalties if they fail to meet this standard.
“Nothing motivates a company more than money but unfortunately I do not see this happening as the regulators and the exchanges tend to act together,” he said.
Infrastructure costs
If the financial penalties were rigorous enough then they could perhaps force the exchanges hand’s by making it economically logical to make the investment necessary to ensure 100% availability. However, this is still at odds with an industry that demands speed and ever changing technology.
Ultimately it is up to exchange’s customers to decide what they value more because it is inevitable that if exchanges are forced to invest in more robust systems then the cost of this will be passed on to them.
While outages at exchanges are dramatic events they must be put in to context. And when they are the buy-side must then consider whether they are willing to pay the extra market infrastructure costs for that extra percent or half-percent of resilience.