Invesco to build proprietary trading algos

Global asset manager Invesco will develop a number of equity trading algorithms internally to execute orders in the US market, but has said the move will not reduce its reliance on the sell-side.

Global asset manager Invesco will develop a number of equity trading algorithms internally to execute orders in the US market, but has said the move will not reduce its reliance on the sell-side.

Saurabh Srivastava, global head of electronic trading for Invesco, told the firm was in the process of developing a number of proprietary algos to sit alongside broker offerings it uses.

In-house algo development, long the mainstay of quant-focused hedge funds, has in recent years interested traditional buy-side firms looking to gain a competitive edge.

The algo tools developed by Invesco will focus on core strategies favoured by asset managers, and will be driven by a team of two quants with experience from the broker and high-frequency trading disciplines.

“Our internal algos for the US market will always be benchmarked against broker algos, covering strategies including implementation shortfall and volume weighted average price. This is a natural extension of the buy-side’s role in algorithmic trading,” he said.

Srivastava, whose firm in December reported US$780 billion in assets under management, did not disclose the exact number of trading algos the firm would develop or when the first proprietary algos would begin executing orders.

According to Srivastava, key US market structure changes would alter traditional broker relationships, although buy-side firms, including Invesco, would continue to rely on a combination of high- and low-touch sell-side services to source liquidity.

Fragmentation caused by a greater number of trading venues and the impact of high-frequency trading, which can impede the buy-side’s ability to execute large orders, have led to asset managers demanding detailed execution data and re-thinking trading strategies used, to improve performance.

“Changes in the market’s structure and its complexity mandate that we foster a different type of partnership with the sell-side than we have historically,” he said.

“Since the use of electronic trading tools to implement investment decisions has become an important part of the implementation process, one of the ways to realise differentiation at that level is to create your own tools,” he said.

To truly reap the benefits of electronic trading, automation and the scope of quantitative metrics, Srivastava said asset managers must increasingly rely on their own resources to shape their execution strategy and foster a greater understanding of how algorithms work.

“To improve execution quality, the buy-side must adapt and develop competence in being able to mathematically model and quantify different phases of the implementation process,” he said.

Skill shift

Commenting on the development, Sang Lee, partner at research consultancy Aite Group, said Invesco’s decision to develop algos in-house was emblematic of broader changes occurring within asset managers.

“It’s a natural progression,” he said, adding that common algo strategies were replicated across brokers, potentially leading to lower execution quality compared to using proprietary tools.

He said although only a small number of large, well-resourced, asset managers could meaningfully develop algos internally, a migration of personnel from the sell-side to the buy-side had shifted key skills to asset managers.

“A lot of sell-side talent has moved to the buy-side, bringing in knowledge of trading strategies and how to build quant teams,” he said, speaking to

Lee said that the combination of greater buy-side experience in algorithmic trading coupled with a push to quantitatively break down transaction cost analysis data had also resulted in asset managers having the ability to develop algos internally.