Building a Better Algo
Traders Magazine Online News, March 20, 2012
Algorithms today are much better at trading illiquid, small-cap stocks than they were in years past. Previously, algos simply sliced up orders and spread them out across time and venues.
Those strategies were largely ineffective at responding to sudden bursts of liquidity. Thats why algos today are taking a different tact.
The algos that the buyside were most familiar with, VWAP or TWAP for example, utilize historical schedules and statistics that may have no resemblance to what is actually happening in the market today, said Dan Nachtman, a principal in global execution services at Bank of America Merrill Lynch.
Liquidity patterns for small-cap stocks can vary so significantly from their statistical averages that it can be difficult for those older strategies to effectively trade small-cap stocks, Nachtman added.
In order to best manage small caps, algo developers today make sure their products recognize when there is liquidity in the market so they can pounce on those opportunities. At the same time, this new breed of algos recognizes when there isnt significant volume, and in those instances they are patient enough to wait for liquidity to return.
To come up with better strategies to trade small caps, software development firm Tethys Technology looked at the global algos it had for trading internationally.
Even some relatively developed markets like Australia and South Africa often have wide spreads and spotty liquidity compared to stocks in the United States. Tethys noted that like U.S. small caps, stocks in other countries tended to have a lot of liquidity all at once, but then see that liquidity quickly disappear.
For us, it was basically taking those techniques which wed learned in those markets and applying them to the U.S. small-cap market, and they worked very well, said Nitin Gambhir, chief executive officer of Tethys.
Sometimes, traders place a large minimum size on an order, which effectively means an algo will have to route the order serially rather than sending it out parallel to multiple venues simultaneously.
Jose Marques, global head of electronic equity trading at Deutsche Bank, said he generally cautions against large minimums, since the algo knows which venues are prone to information leakage and automatically avoids sending them small orders.
The algo is pretty smart, Marques said. Let it figure out which pools its going to and what are minimum interaction sizes for the different pools.
Otherwise, traders are going to end up missing a lot of liquidity that they could have otherwise had, he said.
For more information on related topics, visit the following channels: