Trading: Banks swim in dark pools


Peter Koh
Published on:

Hope to improve prices and minimize fees and impact on market.

In May, Goldman Sachs, Morgan Stanley and UBS announced bilateral agreements to offer reciprocal access for algorithmic trades to each other’s "dark pools".

Dark pools are matching engines for non-displayed, anonymous trading away from the public markets.

The arrangements, which apply to US stocks only, connect three of the largest investment bank-operated dark liquidity pools in the US – Goldman Sachs’s Sigma X, Morgan Stanley’s MS Pool and UBS’s Pin ATS – but stops short of full integration. The banks involved have also left open the possibility that they might connect to other dark pools in the future.

By linking their internal dark pools, the banks hope to improve the proportion of trades executed away from the public market, giving them a chance to offer price improvement, minimize the market impact of trades and save on exchange fees.

"We’re confident that providing our respective clients access to each other’s liquidity will achieve even better crossing results for our clients in an increasingly fragmented market," says Greg Tusar, managing director, Goldman Sachs Electronic Trading.


For the investment banks involved, simplifying the increasingly complicated trading environment was also an objective.

"Our clients are looking for incremental liquidity without having to split each order across many different algorithms," says Will Sterling, managing director, UBS Electronic Trading. "These agreements should offer clients access to additional high-quality liquidity without making their trading process more complex."

The move is another symptom of institutional traders’ frustration with public markets since Reg NMS was introduced and of investment banks’ determination to minimize exchange fees.

"Institutional traders see the public markets as increasingly toxic," says David Easthope, an analyst at consultancy Celent. "The growth of ECNs and algorithmic trading has contributed towards decreasing order sizes over the last decade, making it harder to trade large blocks of stock. There is also a rising risk of market leakage in the public markets as predatory hedge funds study trading activity to sniff out algorithmic trades and trade against them."

According to Easthope, the controversial Reg NMS has exacerbated the trend. "Reg NMS has created a public market designed to protect retail investors and not to serve the needs of institutional traders. As a result, institutions are trying to get as much of their trades done outside of public markets as possible."

Celent estimates that broker- and exchange-sponsored dark pools trade 7% to 10% of the total volume of equities in the US. Depending on the exact definition, however, that figure rises to 12% to 15%. There are estimated to be more than 40 such facilities, with a variety of market models, including many run by independent agency brokerages, up from just a handful two years ago.

By catering specifically to algorithmic trades, the Goldman Sachs, Morgan Stanley and UBS joint approach contrasts with that of some agency brokerage-run platforms including Posit, Pipeline and Liquidnet, whose buy-side clients often want to exclude such liquidity. Some, particularly hedge fund traders, distrust investment banks’ internal dark pools out of a fear of information leakage but most are happy so long as the final execution result is good.

Fight back

A higher proportion of trading upstream from exchanges is of course a threat to their volumes and businesses. Exchanges have, however, been fighting back with innovations such as hidden order types and by linking to dark pools and even creating ones of their own.

Exchanges with this model seek to match incoming orders against liquidity on the order book and then route unmatched orders to dark pools that show indications of interest to the exchanges’ system.

By linking to dark pools the exchanges hope to reassert their position in the execution chain, becoming senders and therefore controllers of dark pool liquidity as well as recipients.