Conventional wisdom says it’s better to give than receive. Instinet’s new algorithm does just that, helping the buyside provide liquidity to the market rather than take it.
The new algorithm, MAKE, is geared toward passive traders and gives them electronic access to the same type of advanced liquidity-providing tactics used by market makers. Using historical information, recent trading patterns and real-time market data, MAKE sizes and distributes child orders at multiple price levels and across destinations, controlling adverse and negative selection while reducing exposure to signaling, gaming and predatory techniques.
MAKE’s genesis stems from its logic, which was originally developed for liquidity-providing in the other algos in the agency-only broker’s Execution Experts suite. Because of customers’ interest in a strategy that only provides liquidity, in much the same way an electronic market maker does, the broker created the separate MAKE algo.
MAKE is available in all 37 markets served by Instinet electronically across the Americas, EMEA and Asia-Pacific, John Comerford, head of global trading research at Instinet, told Traders. What makes MAKE unique in the algo world is that it is designed to only trade passively, never taking liquidity, which is generally a much harder thing to do effectively than simply hitting bids or lifting offers in the market.
When taking liquidity from the marketplace, an algorithm has control of when, at what price, on what venue and at what size to trade. Depending on a trader’s goals, the algo can also react immediately if opportunities for attractive prices or larger prints, Comerford said.
But in the case of providing liquidity, the timing of trades cannot be directly controlled, so the algo must manage posting decisions carefully. Factors such as how much of the order to expose, on which destinations and at what price level must be considered to balance negative selection (missing potential liquidity at a “good” price) and adverse selection (providing too much liquidity and limiting price improvement). Market conditions must also be continually monitored to try to detect liquidity imbalances, allowing the algo to pull back from the market, or add more liquidity, as appropriate.
MAKE’s trading logic are controlled by a simple yet highly customizable control set, allowing users to specify a range of different styles of liquidity providing, from “patient” to “urgent.” At lower urgency levels, Instinet’s signal-driven FADE logic will be more sensitive to potentially adverse market conditions, and will briefly pause or slow trading to reduce adverse selection. At higher urgency levels, with more emphasis on minimizing negative selection, child orders will tend be larger and more frequent.
“The algorithm really does it all. The logic is designed to watch the market constantly, manage the order queues and might cancel an outstanding child order,” Comerford said. Depending on the liquidity and idiosyncrasies of a particular stock, MAKE can pause an order from a few milliseconds to as long as a minute, he added.
Comerford also said MAKE is particularly well suited for low-information trades where opportunity cost is less important than impact reduction-particularly at low or moderate urgency levels-or large orders, where limiting information leakage and potential signaling is paramount. At higher urgency levels, it can also be used for trades with higher opportunity costs, trading larger volumes while remaining passively priced.
“The tool is tunable-so the user can easily select the way it provides liquidity,” he said. “For users seeking to better manage their passive side trading, MAKE provides the opportunity to realize significant spread savings while simultaneously reducing the costs and risks associated with liquidity providing.”
Instinet, the execution services arm of Nomura Group, has made its MAKE algorithm part of its Execution Experts algorithmic trading suite.
(c) 2014 Traders Magazine and SourceMedia, Inc. All Rights Reserved.
http://www.tradersmagazine.com http://www.sourcemedia.com/
