BestEx Research launched its AMS One platform on June 9. Traders Magazine sat down for a candid conversation with Hitesh Mittal, Founder and CEO, to talk about why competing on execution has been so hard for all but the top few firms, why it requires a focus on market structure and research, and how AMS One solves the problem.

What is AMS One?

AMS One is a SaaS platform that helps banks and brokers launch or upgrade their algorithmic trading business across global equities and futures markets. The platform comes with our execution algorithms, which offer highly competitive performance versus tier 1 banks, and an algo customization tool that an electronic trading desk can use to build and customize their own algorithms. But running a successful execution business requires more than that; this platform is an end-to-end solution that offers not just algorithms but all the tools to manage and service them. For example, it comes with an EMS and a TCA platform which create a seamless experience for electronic trading desks and quants. The platform is fully managed, which brings infrastructure cost down to zero. Clients can launch their business in a matter of weeks or  expand their current regional business to other markets easily.

Algorithmic trading has been around a long time, and you built and managed platforms for the buy side and the sell side before founding BestEx Research. What has changed since then, and what hasn’t?

I built the global equities algo product at ITG, which was really a pioneer in the field, then moved to the buy side as head of advanced systematic trading at AQR where I built the firm’s multi-asset algorithms in house. What hasn’t changed is that great execution is still hard to deliver across global, multi-asset markets. It takes constant research and a close eye on changing market structure. What has changed is the bar for quality. A functional set of algorithms used to be enough to compete, but today it isn’t. Most large buy-side firms have hired quants who measure and reward brokerage firms on performance and many of them also demand bespoke solutions. That trend started in US equities and is spreading globally and across asset classes. Execution today is half about algorithm quality and half about the customization and consultation wrapped around it. 

How is the competitive landscape changing as buy-side firms focus more on execution cost?

Except for the top few banks most others do not have a competitive global equity offering. And most banks, including the top tier, do not have a competitive futures platform. Not being able to offer a competitive low-touch trading solution is hurting not just their execution revenues but also other aspects of the business such as prime brokerage, research, central risk books and market making, all of which have natural synergies with the execution business.

When it comes to the global aspect of an execution offering, there is a major volume and complexity mismatch that affects investment. US equities is the largest market in the world, so it was a relatively easier decision for banks to justify building algo products in the US. But EMEA and APAC markets see only a fraction of US equity volume, while having more complexity and much higher market data and infrastructure costs. Those markets remain underserved from a product quality perspective, presenting both an opportunity for those who do invest and also a challenge due to the tradeoff between volume and complexity.

How are the global and regional banks and brokerage firms currently participating in this space given the challenges you’ve outlined?

We see a combination of things. For equities, some are white labeling other banks’ algorithms. Some are building upon regional software vendors’ and EMS platforms’ offerings. Many firms who invested in the early cycle of algorithmic trading have a home grown legacy platform in one market but use vendor or white label solutions in others.

For futures, there is little to no investment in algorithmic execution and most banks are using their OMS vendor’s solution or using more of a copy-paste approach from their US equity solution, which is suboptimal given the structure of futures contracts themselves and the differences in market structure.

The issues with white labeling another provider’s algorithms are pretty severe. You have no transparency into the algorithms or the ability to customize. TCA capabilities are limited. While this may have been ok 10 years ago, it’s not anymore. The real choices most sell-side firms have are either to build a solution in house or to use a vendor solution such as AMS One.

What problems does AMS One solve that other vendors don’t?

There are a few key elements here. First, I keep going back to two things you need to build a successful algorithmic trading business—high-performance algorithms and the ability to customize them and service your clients. Most vendors’ solutions can offer you the service tools but the algorithms they come with off-the-shelf are simply built for buy-side firms who don’t measure the costs. We’ve heard some complain about spending years to get the basic functioning of vendor algorithms right. But BestEx Research is an algorithmic trading firm first and a software vendor second. Our algorithms go head-to-head with tier 1 firms from day one.

Second, AMS One is the only global equity algorithmic trading solution for sell-side firms and the only global futures solution that isn’t just a feature just bolted on to an OMS. After the success of our equity and futures algorithms in North America, we spent the last four years building a complete global suite, market by market, optimizing for each market’s idiosyncrasies. And for our sell-side clients, that means stepping into a new market is as simple as turning it on.

Last is our service model. We are fully managed which means that banks will spend less for an AMS One subscription than they would spend just managing another vendors’ infrastructure. Our support desk doesn’t look like a help desk, it looks like a trading desk. We proactively monitor trading, and we’re there on Bloomberg, phone and Slack, not buried behind a ticket queue. 

Put those three things together, and you have something rare in AMS One. It lowers costs and grows revenues at the same time. That’s a value proposition that is hard to find. 

What about teams who already have quants building or maintaining their own algo platform? Is AMS One still appropriate for them?

Absolutely. I get asked this question all the time. You may find it surprising but banks who have quant resources and invest in strong electronic trading desks are some of our most successful clients. That’s because with our platform they are no longer stuck on maintenance. No quant wants to spend their time cleaning data or fitting a volume curve for an index rebalancing day, and desk personnel don’t want to figure out how to connect to every new venue that comes along. We take care of those critical details, which frees the team to add measurable value for clients—actionable TCA, strategy customization, real consulting. Our platform allows them to use their skills to grow revenues, and puts them on offense  rather than on defense.

How was BestEx Research able to build a global, multi-asset solution in such a short amount of time?

We are actually nine years old now, with 75 people with 20 quants, and most of the rest being software developers or in client consulting roles. We continue to grow every year both in terms of trading volumes and head count. Last year, we traded more than a trillion dollars in volume.

But to get back to your question, we can expand quickly because it’s the way our software was built. AMS One was purpose-built for algorithmic trading as a low-latency, lightweight platform in C++, and building a multi-asset global platform was the vision from day one. TCA, simulation, EMS, algos, and trading analytics all live inside AMS One as opposed to being built separately. It took us three times as long to build it this way than if we had latched it on to an EMS but this foundation has allowed us to scale more easily.