Sometimes, less is more.
It was almost 20 years ago that the first algorithm appeared in the equities market. The VWAP algo was a simple electronic trading tool for a simple time. Over the years, as market structure changed, so did the algorithms that traders needed tfor their trades. Simple VWAP and TWAP algos morphed into more complex varieties, such as implementation-shortfall and liquidity-seeking types. Then came tools designed specifically for operating in dark pools, those that searched actively and passively, as well as those that siffed out so-called predatory traders.
As the buyside made more demands of its algos, brokers and vendors alike flooded institutional trading desks, with algos surpassing the number of brokers listed on a traders desktop. And they came with names such as Chewbacca I, Chewbacca II and Chewbacca III-all similar but yet different in minute ways. And then the buyside complained that there were too many algorithms, which made a traders job harder and, negating the purpose of the formula in the first place.
And thats just in equities. Algos sprouted for options and currencies, too, all targeted at the growing needs of the buyside trader who now executed orders in multiple asset classes throughout the day.
So, is there a return to fewer algorithms that are simpler for the buyside trader to use?
Cheryl Cargie, head trader at Ariel Investments in Chicago, said that she and her firm are being pitched fewer algos these days.
I think the algo market is pretty saturated at present, Cargie began. Everybody on the Street has one, and in my opinion theres nothing new out there that hasnt already been done or put into algorithmic form.
David Margulies, head of the electronic trading group at Weeden, said that his firm saw the move to bespoke algorithm development several years ago and is now seeing the trend wholesale in the equity markets.
Without a doubt, we are seeing more bespoke algorithms than ever before, Margulies said. The buyside has gotten a lot more knowledgeable with order types and venues, and not just relying on the brokers suggestions. They are telling us exactly what they want, and giving us their ideas and thoughts with regards to execution, making development of an algorithm more of a consultation.
For Weeden, the buyside trader discloses his individual trading style-execution venue preference, order type, etc.-and Weeden in turn will incorporate these factors into a cusom algorithm.
Were doing more customization than ever before, from making minor tweaks to wholesale changes, Margulies said. We go back and forth with clients and customize the algo.
By making its tool design architecture very open and flexible, Weedens development time has shrunk from months to, in some cases, days.
Our development process is like Lego blocks; we can just string together the right components and build them into an algorithm, Margulies said. Theres virtually nothing built from scratch anymore, and thus lead time has shrunk dramatically.
