Linette Lopez, BI, The Number That Has Goldman Sachs Worried About High Frequency Trading, here. Aren’t these the folks who wrote their core HFT high performance code in Erlang? How’s that working out?
So once big institutional clients — the mutual funds and hedge funds that HFT firms love to pick off when they notice the institutionals’ big block trades in the market — started complaining about HFT, Goldman knew it was time to change their strategy.
Patterson and Baer reported that at a meeting in London several weeks ago, Goldman’s institutional clients voiced concerns that are now familiar thanks to Michael Lewis’ book, ‘Flash Boys‘. They said that they felt HFT firms were given an unfair advantage and that the market was too opaque, complicated and dangerous.
That’s when Goldman started sending around internal memos asking for commentary on market structure, and COO Gary Cohn wrote the anti-HFT op-ed that shocked people across the Street.
In the op-ed, he mentions one more issue that has Goldman worried about HFT. The bank is known for having some of the best technology in finance, but last August a glitch in its software sent erroneous quotes into the market and cost the bank $100 million. And Goldman doesn’t lose $100 million.
The economic model of the exchanges, as shaped by regulation, is oriented around market volume. Volume generates price discovery and liquidity, which are clearly beneficial. But the industry must recognize how certain activities related to volume can place stress on a market infrastructure ill-equipped to deal with it.
In other words, exchange software is now so complicated that it is not something a firm can do as a side show — it has to be the main event.
According to industry analysis, since 2005 the flow of these order instructions sent through U.S. stock exchanges has increased more than 1000%, yet trade volume has increased by only 50%. One consequence of the enormous growth in order-message traffic is that increasingly the quote that an investor sees isn’t the price he or she can transact, as orders often get canceled at lightning-quick speeds.Currently there is no cost to market participants who generate excessive order-message traffic. One idea would be to consider if regulatory fees applied on the basis of extreme message traffic—rather than executions alone—are appropriate and would enhance the underlying strength and resiliency of the system. Regulators in Canada and Australia have adopted this approach.