French investment bank Société Générale has announced the details of Eclipse, its new liquidity-seeking algorithm, which analyses the quality of liquidity that can be found in dark pools.
The algorithm, which is available in Europe and is in the final stages of development in the US, hunts dark and lit liquidity across a variety of venues including multilateral trading facilities, exchanges and broker-owned crossing engines. As brokers do not generally allow third-party access to their crossing networks, SocGen is required to become a client of the brokers whose pools it wants to access.
According to Mark Goodman, head of quantitative electronic services, Europe at SocGen, one of the main differentiators of Eclipse, compared to other liquidity-seeking strategies, is its approach to measuring the quality of dark liquidity.
“Clients have told us that there is currently little analytical distinction between the dark venues that are available, with most algorithms adjusting their dark pool participation based on anecdotal evidence,” Goodman told theTRADEnews.com. “We developed a Quality of Venue Measure to look at the level of information leakage and adverse selection from each dark pool, and designed the logic for Eclipse based on our results.”
For example, one measure used looks at the volatility of the mid-price of a stock on a displayed exchange, after it has been traded in the dark. If the price of a stock moves adversely in the lit following a dark execution on a regular basis, one can infer that the trading counterparty is looking for information, rather than liquidity.
Using this type of analysis, clients can adjust the behaviour of Eclipse to suit their trading strategy, e.g. by posting larger portions of orders in those dark pools that are considered safe, or excluding those venues that have the most information leakage.
Goodman added that analysis of the liquidity quality is reviewed on a monthly basis, as new participants join dark pools and existing members adjust their dark trading strategies.
He also observed that the liquidity measurements used threw up some unexpected results.
“There were a few venues that turned out to have higher levels of information leakage than we expected,” said Goodman. “This taught us that even if you know the characteristics of a venue and its philosophy, you need to have a systematic and analytical way of measuring performance.”