Remember me

Register  |   Lost password?

The Trading Mesh

High Frequency Trading Solutions from Mantara

Mon, 20 Jun 2011 09:29:00 GMT           

Mantara is a name that seems to have been popping up again and again recently in the context of high frequency trading solutions. The firm recently announced the opening of its Toronto office, to support what it sees as a growing demand for its HFT solutions in the Canadian market.

Not knowing much about the firm, I was curious to find out more, so I spoke with Michael Chin, President and CEO, and Shawn Sloves, Co-Founder and Head of Product Management and Strategy at Mantara.

HFT Review: Michael, I’ve seen your firm in the news a couple of times recently. You received a $7m round of funding in May and just opened your first international office in Canada. These seem to be exciting times for Mantara.

Michael Chin: Yes they are.

HFTR:   Can you give me some background on the firm and where you position yourselves within the HFT “ecosystem”?

MC: Sure. In terms of history, Shawn dates back to the very beginnings of Mantara.  I came on board in September 2010, when the decision was made to regroup and define a strategy to accelerate revenue growth. We had an incredible product that had been three years in the making, with engineers going almost to the point of perfectionist in building out Shawn’s vision of a high-performance trading platform spanning across multiple asset classes. Thankfully Shawn and others here had the foresight to think about risk on a pre-trade level from the very beginning and build it into the “DNA” of the platform.

When I came on board, I was tasked with expanding Mantara’s footprint across a much broader group of market participants, which we’ve been able to do quite successfully. I didn’t do this on my own by any means; I was fortunate enough to bring together a client-facing organization with over 100 years of relevant experience in the markets.  So it was a kind of “plug and play” team, from sales people and account managers, to implementation managers and client services and support people. We basically invaded Mantara and along with Shawn, launched a much more aggressive strategy to get this product out to clients.

HFTR: And that approach has yielded results?

MC: Yes, really good results. We’ve tripled the number of clients on the system in the past six months and—more importantly—we have experienced double-digit growth with regards to monthly reoccurring revenues. You’re going to see a lot more about Mantara coming on to the market because we have that many more clients. And we’re really leading the charge with our pre-trade risk offering, both in its performance, but I think more interestingly in terms of how we’re deploying it and how we’re thinking about risk on an aggregate level. We’re also building out our reach beyond the U.S. much more aggressively.  As you know, we just opened an office in Toronto and we’re scouting out space from a data center perspective, in both the UK and in Japan.

HFTR: Tripling your client base in six months is pretty impressive. What type of clients are you typically bringing on board?

MC: First are end-user clients, the buy sides, in particular those looking for ultra-low latency levels of performance. The whole space is changing rapidly as the need for this type of product expands beyond the traditional high-frequency trading firms and “black-box” traders to the standard hedge fund, buy-side clients. They’re moving in the direction of being less reliant on broker algo offerings and are instead incorporating their own tools to build their own algos and take ownership of the trade and in doing that, not sending those trades through the traditional smart order routers of the brokers, but instead using tools like Mantara’s for sponsored access capabilities. They’re looking for exchange connections, using the broker’s MPID. This is exactly Mantara’s business.

HFTR: So you’re facilitating broker neutrality and disintermediation?

MC: Well, clients can still pay their brokers for research and for the MPIDs that they use, but they’re doing analysis on the value they’re receiving from the broker algorithms being offered versus the commissions that they’re paying, and they’re seeing that in many cases, the broker algos don’t perform as well as the brokers claim. Some of these broker algos are simply a tool for the broker to make money on the rebate game, or to fuel their internal crossing to get information and actually use it against the clients. So we’re seeing a lot of change in the mindset of the buy side.

Then there are the brokers themselves, who look at Mantara in two ways.  The first is as an in-house solution, where they take Mantara software, including our market data ticker plant, exchange connectivity, order routing and pre-trade risk modules, and incorporate it into their data centers to replace archaic older platforms that are just not at the level of performance they need to compete in this market.  So we have gained a lot of traction in this area.  We have two very large deals where we are essentially the in-house solution for these brokers and we run all their risk, their order routing and their market data for them.

The other way that brokers use us is as an outsourced solution to provide market access and pre-trade risk to clients that they want to come and trade with them, via sponsored access. Brokers can offer Mantara as an “in-the-box” solution, so that their clients have the technology end product to be able to build their own black-box models.  Their clients can get market data into those black-box models, they can run pre-trade risk, and get out to the market through our order gateways.

Finally, you have the exchanges who are now being required to put mandatory risk checks in place.  A number of exchanges are looking at Mantara either as a solution to offer pre-trade risk checks at the exchange level or to offer our product as an additional upsell service to clients who want to take the exchange’s risk offering as an additional product.

So these are the three main client types that we’re focused on. On the exchange side, where we had no business before, we’re now doing business with one of the very large exchanges and we have a few more in the pipeline.  On the broker side and the buy side, we’re seeing a considerable amount of growth.

And we expect to see more growth once the market access rule is in place.  Then of course there is growth to the other regions like Canada, where they’ve come out with their own ruling similar to the SEC’s, and MiFID II in Europe has a lot of discussion points around pre-trade risk and HFT.

HFTR:   By trying to cover all these different types of firms, buy sides, sell sides, exchanges and so on, isn’t there a danger of you ending up with a lack of focus by trying to be all things to all men?

Shawn Sloves: No, because it’s all the same service. We’re offering a holistic view of risk. If you think about the end users, there are different levels of risk checks that you need to do on a pre-trade basis.  From the exchange’s perspective, the exchange has to manage the risk of the brokers who are member firms, so they’ll look at the brokers MPIDs being used to access the exchange. They’ll typically do a “fat finger” check or look at total shares or dollar values across all the aggregate flow coming in through the broker’s ID.

Then you’ve got the brokers. They manage client risk so what they have to do is know the clients on an individual basis and set up their own risk checks like fat fingers, total dollar values and other account-level risks that involve margin for their customers.

So the exchanges manage the brokers and the brokers manage the clients.

Now the clients normally don’t trade with just one broker at the sponsored access level, they’ll typically use multiple brokers.  If I’m a hedge fund and I’m using three major brokers on a pre-trade risk basis, I want to see an overall aggregate risk across all three brokers, and we facilitate that.

At Mantara, we have a holistic view on the market.  You have the exchange managing the risk of the brokers, the sponsoring brokers managing the clients’ risk and the clients managing their own risk across multiple brokers.  The checks are all the same, it’s just the permissioning and the way the system is used by the end user that is different. The mechanism and how they use it varies slightly but the logic is still the same.

HFTR: Which markets and asset classes does it cover?

MC: We have grown the asset classes on the basis of the demands we’re seeing for low-latency products. US cash equities is the obvious one, US equity options is another asset class we currently support.  The next on our road map is US futures, mainly index futures.

HFTR: In terms of connectivity into those markets, do you go in via the exchanges’ native APIs or do you offer FIX connectivity, or is it a combination of both?

MC: It’s a combination of both, whichever is the fastest way to get to that exchange.  Going into our gateway, a client has the ability to go native or FIX.  We have our binary API, which is basically the way that our system communicates across all its modules, and then there is FIX, which has a layer of translation so it’s going to be slightly slower.

HFTR: Focusing on the pre-trade risk element, is that done in software or do you use any programmable hardware like FPGAs?

MC: We have chosen to build it on a smartNIC card instead of on FPGA. There are some great advantages to this approach, which I’ll let Shawn talk about it.

SS: The primary difference between the FPGA and our smart card is that FPGA requires very low-level language programming, whereas with these smart cards, you can write code in C or C++.  So whatever we run our services on, software or commoditized hardware, that same software will run on one of the smart cards.  There’s a unique processor that runs a Linux OS, which allows us to execute our software embedded on the card.  It’s flexible, we can change it very quickly and we write code much like we do for servers, so there’s a far quicker time to market.  And we believe that this smart card solution offers speed advantages that are comparable to FPGAs.

MC: We actually have speed advantages that are better than FPGA when you look at the full implementation cycle, because the whole concept behind the smartNIC – is way faster. The great thing about the smartNIC solution is that it fits into the network slot that already exists on the client’s machine. It’s like running a server within a server, unlike FPGA where you have to run an FPGA solution on a separate server from the client’s box.

People talk about the speed of FPGA but what they don’t tell you is that in order to get to the FPGA, you actually have to go across the network, go in and go back out again. Even under the best conditions, under the best optimized network and switches etc., we estimate you’re going to add a minimum of six to eight more microseconds to the entire process.  Whereas by inserting the smartNIC right into the client’s server, you only incur 300 nanoseconds in terms of communication between the client’s strategy and the smartNIC card. So we’re talking a total of four or five microseconds to run 12 risk checks on our card versus anywhere from eight to ten microseconds to run the full feature of an FPGA, which is maybe 3 or 4 risk checks.

HFTR: In terms of deployment, do you offer your solutions both in-house and on an ASP model?

MC: Yes, we offer both models. We have our own co-location cages at all the major data centers in the US: Carteret, Mahwah, Savvis and Equinix.  We also offer proximity hosting in our own data centers here in our offices at 111 Town Square Place in Jersey City, and clients then have options to have multiple co-lo strategies using Mantara’s cages. So there is a lot of flexibility on how clients can choose to deploy our solutions.

HFTR: Can you give me any kind of indication on pricing?

MC: What I can tell you is that we offer a much more competitive product, price-wise, compared to FPGA. FPGA-based solutions, with the time to market and the efforts & resources necessary to make any changes and so on, the cost of FPGA can be in the neighborhood of $50,000 per month, whereas our smartNIC product is a fraction of that.  In addition, we price on a fixed license basis. Unlike some of our competitors, we’re not in the business to compete with brokers.  We’re not looking for the variable pricing model.

HFTR: To wrap things up, can you tell me what your plans are for Europe?

MC: Yes, we actually have two very large global investment banks that have successfully rolled out our product here in the US markets, and we’ll be building out to the European markets with them in Q3 this year.  Independent of that, we are looking at data center space to allow us to continue with our own co-location strategy for clients, so we’re scouting out space for that now in London and also in Tokyo, because we have some clients that are pushing us out to Asia.

HFTR: Great, thanks Mike, Shawn.

 

, , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,