Algo Consolidation Speeding Up

Fewer is better. That’s the message big brokers have been hearing from their institutional clients over the last year, as buyside traders have been rationalizing the number of algorithms on their desktop and ridding themselves of the algos they don’t use.

This move to keep only the best performing algos has sped up of late, as volatility and volumes have skyrocketed, and also because buyside traders don’t want to sift through a list of algos that are simply taking up space.

"As a trader, you reach an inflection point where it is intellectually impossible to evaluate, understand, and measure the real-time performance of all your different strategies, and identify the appropriate scenario to apply them," said Tim Reilly, head of North America electronic execution sales at Citi.

About two years ago, that led to a movement to simplify the front end and the order ticket to create more efficiency in work flow and strategy selection, Reilly said. With the recent increase in volatility, change has become more rapid.

In fact, Reilly sees the period of July 25 to the present as a radically different execution environment from the beginning of the year. Clients want to streamline for the new situation, but at the same time they are looking for increased customization.

"We’re seeing a much more active engagement from clients around how to best utilize their electronic toolkit," Reilly said.

And that doesn’t just mean using the algorithmic tool kit in isolation. Reilly said buysiders are evaluating all the different options they have–capital, sales trader, electronic trading toolkit, crossing network–and finding a suitable map for the best execution for that order.

Since July 25, Citi has found its average order size is up by more than a third in participating algos. Meanwhile the average duration of an order has declined by more than 10 percent. Due to the explosion of volume, algos are trading larger orders, faster.

Reilly said some clients are trading too rapidly, and they’re creating an impact on the market. He said Citi is working closely with clients to encourage them to adjust the strategies they select so they can achieve the best trading outcomes.

Bank of America Merrill Lynch has gone so far as to consolidate its algo platform globally. The firm’s new Instinct product allows clients to fine-tune parameters on an order-specific basis, choosing a level of urgency from one to five.

The firm rolled out Instinct just in time for the massive volatility this August. That provided for a natural stress-test of the system, and the company says adoption of the new product his been high, and it is now its most used algo.

According to Michael Lynch, head of Americas electronic trading for BofA Merrill, Instinct simplifies the front end and puts the complexity on the back end. It’s designed to work with all levels of capitalization, but because it is signal based, the benefit is most noticeable in stocks where there is an inconsistent volume pattern, he said.

"You can mitigate your signal risk by responding opportunistically to liquidity events," Lynch said. "With Instinct, this adaptability minimizes order information leakage in the marketplace."

Lee Morakis, head of execution services sales at BofA Merrill, said as algo consolidation continues, he could see clients cut back to using only four or five different algos. That might not happen for another year, though, as traders get their technology in line with where it needs to be, he said.

Regardless of how few algos remain standing, most people in the industry believe the time has been ripe for consolidation.

"If you’re a trader sitting at a desktop and if you have five algo providers with eight to 10 strategies each, very quickly you get up to 50 different tactics on your pull-down menu," said Jose Marques, global head of electronic equity trading for Deutsche Bank.

Marques said he sees a lot of buyside firms picking and choosing the best of breed algos in different categories and then creating a single button for each which hides the broker name from the front-end trader. He said this is done to simplify workflow more than anything else.

While consolidation has been going on for a while, Marques said it has been picking up recently, and some front-end vendors are providing technology that makes it easier to do.

Another big trend he’s been seeing is the uptick in opportunistic algos that aren’t tied to a benchmark like volume-weighted average price. He said a firm can build an algo that gets them VWAP every time, but from a total-return perspective, there might be a cheaper way of implementing the trade.

As the market has gotten volatile, the natural response is to shorten trading horizons, and while there has been some of that, Marques said there has also been more hand management of trades. Buyside traders are still using electronic tools, but being more dynamic in deciding when to go ahead with an order and when to pull back for a bit.

"When the market gets volatile, it’s harder, because things are moving fast and you’re busier, but it’s also a time when a human trader on the buyside can potentially add more value," Marques said.

While it is impossible to predict how long the volatility will last, Marques said it doesn’t look like the macro events driving volatility are going away in a hurry.

In that environment, algos have to be self-adaptive to volatility regimens, according to Marques, and that means building an algo from the ground up to take account for different levels of volatility.

"If you built an algo that did a great job when the VIX was at 15, and now it’s a 40 vol world, you’re not going to see the kinds of results you want to see unless your algos are very dynamic," Marques said. "The world has moved on from static heuristic-based algos."

Today, algos are getting smarter and more aware of the market environment, he said. And that’s leading to better trading outcomes, in spite of the decline in the number of algos.