Tools that help traders reach many dark pools at once can be a detriment to best execution, according to a recent study.
Using "dark pool aggregators" can actually work against you. In fact, they can increase your chances of incurring adverse selection on a trade, where the trade is about to move against you or already has. The institutional electronic brokerage Pipeline Trading Systems and the buyside firm AllianceBernstein reached these conclusions recently after conducting a study on the subject. It’s entitled: Adverse Selection vs. Opportunistic Savings in Dark Aggregators.
Adverse selection means selecting the wrong counterparty. Market makers prefer to deal with "uninformed" traders, such as retail customers. They don’t want to buy stocks from "informed" traders, such as sophisticated day traders or hedge funds, who know everything about where a stock is going. For example, a dealer does not want to pay $40 per share to accommodate a well-informed seller for fear that the stock will then drop to $30.
Brokerages have been building algos that access multiple dark pools over the past couple of years as solutions to market fragmentation that so many of the pools have helped create. Different aggregators would let traders access an ever increasing number of pools at once.
And as dark pool volume climbs, aggregator use becomes more significant. Average daily volume traded in dark pools stands at roughly 8 to 10 percent of the overall market, according to many estimates. But as a percent of the total market volume, participation rates in the dark aggregators–where one trades consistently using them–commonly range between 15 and 25 percent, depending on the aggregator, according to Henri Waelbroeck, vice president and director of research at Pipeline.
More than dark pool volume trends, problems of adverse selection can develop from the company you keep. As traders get more of their volume done in dark pools that provide small fills, they interact more frequently with high-frequency trading firms, Waelbroeck said. Trouble arises when high-frequency traders–which represent around 70 percent of overall volume today–use their short-term alpha models to trade at advantageous prices, he added.
"So, as you interact with these guys, if you don’t have any kind of control for participations rates [in aggregators], you’re exposing yourself to the natural adverse selection," Waelbroeck said, "in that, when you are putting out buy orders, then they tend to execute faster as the stock is about to go down, and more slowly as the stock is about to go up."
But the study also reported that it’s possible to both measure and mitigate the severity of adverse selection incurred from using dark pool aggregators. And it said there are tactics traders can apply to use aggregators and reduce adverse selection costs, Waelbroeck said.
"What we really showed in this paper is that you can develop a methodology to measure adverse selection costs, and also measure the opportunistic savings that can be achieved by using dark aggregators," Waelbroeck said. "By measuring these separately, you can identify opportunities for reducing adverse selection costs, selectively, without having to reduce opportunistic savings."
One trader who uses algorithms that access multiple dark pools agreed with the study’s assessment of the risks associated with dark pool aggregators. Whenever his firm uses the aggregators, it watches for any kind of signaling risk.
"In other words, we don’t want to send [an order] to an aggregator, find that we receive a couple of executions, and see adverse market reaction because we’ve essentially signaled a particular buyer," he said. "We probe around in them, depending on how effective they seem to work for the particular orders we’re handling."
Any risk of adverse selection through using aggregators often depends on the liquidity of the stock in the order, he said. A trade of a stock that trades easily won’t likely run into trouble in an aggregator. With less liquid names, though, the risk of signaling your intentions to others increases, he said. This working knowledge informs how he uses aggregators.
"We may just place an order in there, wait a minute or two and withdraw it," he said. "When you leave resting orders in those situations, I think they’re discover-able, given enough time."
Pipeline said in the study it has a solution to the adverse selection problem. The firm built an algorithm switching engine that can be used to shift in and out of "opportunistic strategies" at different times throughout the trade.
Based on the Pipeline/AllianceBernstein study findings, the study recommended the following actions to avoid adverse selection:
* Avoid using passive strategies or manually managing tactical limits early in the trade, especially in high volatility markets when trading non-discretionary orders.
* Use dark aggregators tactically with limited exposure times.
* When using opportunistic algorithms or trading tools, it is important to control the participation rate. Ideally this can achieved preemptively through predictive analytics, but ex-post control measures–like imposing minimum and maximum participation rates–already lead to significant improvements.
* Exploit the alpha gleaned from the participation rate anomaly. On a rise in the dark fill rate, consider pulling back to passive strategies; when dark aggregators run dry consider engaging in the displayed markets to keep to a reasonable schedule.
* Use opportunistic algorithms or dark aggregators in situations with low adverse information risk.
