Industry Spotlight: Anova Financial Networks CEO Michael Persico

April 8th, 2019 by · 1 Comment

The financial vertical has long been a crucial part of the infrastructure ecosystem.  The rise of ultra-low latency trading provided early opportunities for differentiation in what had been commoditized bandwidth and services after the dot com bubble.  First with fiber and then with microwave, one of the specialists that has driven innovation in this segment over the last decade is Anova Financial Networks. With us today to give his perspective on the financial networking business and Anova’s place within it is Anova’s founder and CEO, Michael Persico.

TR: How did you get started with the financial networking and data business?

MP: This is actually my third fintech entity; the first two I built and sold. I go all the way back to 1999 when the markets were just going electronic. My company then had a hand in progressing the markets from open outcry to the electronic trading that it is now.  First there was the floor, and then it moved to the screen. But the screen still wasn't the algorithm at that point, it was just people clicking with a mouse. There was still a human element. There was no inter-exchange arbitrage back then. People were arbitraging the floor with the screens up in their office. Ultimately, things gave way to machines trading, of course.

TR: How did Anova come to be?

MP: FinTech is like The Godfather III. You try to leave, and they just pull you back in. After I had sold the second company, I was just sitting on my couch when a previous customer from the Sydney Futures Exchange called me up and said they wanted to ‘come back home’. So I founded Anova with the Sydney Futures Exchange, which is now ASX, as our anchor client, and it grew from there, starting with legacy services supporting exchanges. But I really wanted to understand what the market was looking for, and to me it had evolved into exchange-to-exchange connectivity, whether it was regional, such as connecting NYSE to Nasdaq, or international, such as connecting the CME to the JPX in Tokyo. So we created this concept of an exchange-to-exchange connectivity provider. That was what the company was founded on. We evolved a couple of times from there, but in 2009 that was our flag in the sand - to shorten the distance between exchanges. At the time we were planning on doing it with fiber, and that evolved as well obviously. February was our 10-year anniversary.

TR: What does your infrastructure look like today?  How do you balance building versus leasing?

MP: We want to own our assets. There's lots of people out there that buy a contract low and sell it high. We call that paper swapping. There's a market for that. But when we were mainly building fiber networks, we got our CLEC license in New Jersey and in Illinois. We would micro-build strategic aspects to a fiber line. Maybe the shortest path is to use some fiber going down Route 95 in New Jersey. But if you can get better private rights-of-way, then that is a differentiated route and it's something that we thought people would have an interest in. When we first started, the routes were 100% leased. Then, as we started adding our custom segments, they became 60% leased, then 40% leased, and eventually as much as it made sense to build and own yourself, because some pieces are just commoditized. But we don't need to reinvent the wheel just to say we own 100% of it. Sometimes there's business sense to using things that are already best in class. But we own duct, we own fiber, we light it, and we own the optronics at the end points. It’s the same thing on the wireless side. We even have our own tower climbing crew for support, maintenance, and builds. And that's a real differentiator. We don't even use contractors there. These are W-2 folks that are on our payroll. We own the equipment. We own the leases. We also actually own our own radio platform. We have FPGA developers where we write our own layer 1 and layer 2 code to handle fair bandwidth management and the termination devices that sit at the endpoint. We don't even use Arista or Cisco; we use our own devices where we've written FPGA code to handle chopping up the bandwidth and queuing and certain aspects of handling the packets. That's a real differentiator for our clients, as well as from the other vendors that are trying to do this service that are buying off-the-shelf equipment.

TR: What geographies does your network focus on today?

MP: Our growth is like that of an airline. We started out regional, then we went national, and finally now we've added international routes. When we started out, our infrastructure was specific to just Chicago and New Jersey. Then we connected Chicago to New Jersey with our own infrastructure. Then we went down to Washington DC from New Jersey and back to Chicago. We're the only provider that owns our own assets across all of the major macro asset classes in the US: equities in New Jersey, futures in Chicago, and events coming out of DC. I'm taking liberties here and counting events as an asset class, but the point is we have wireless assets connecting all three. Finally, we recently expanded to Asia with assets from Tokyo to the landing station. From there we use a subsea provider to go to four locations: Shanghai, Singapore, Hong Kong, and Taiwan. At each of those endpoints there's also custom infrastructure. It is clearly a hybrid network because you can’t do wireless the whole way, and it clearly uses some leased assets because we don't have a transpacific cable. It is at the endpoints that we add the Anova secret sauce or expertise with wireless in Japan and custom fiber in the other locales.

TR: How has the financial community’s use of ultra-low latency connectivity evolved over the years, and how do you approach meeting those needs?

MP: We view latency in the form of a pyramid. At the top of the pyramid are the latency critical folks who will pay a premium for the last nanosecond. At that tier, there's been consolidation over the years and some have exited. Those are people that you can serve, but that's not much of a business. If there were 12 three years ago, there are six or seven right now. But then there's the second tier, which is what I'll call ‘latency important’. They don't need to be first in the queue, maybe happy being in the top quartile. That has been a large growth segment for us. But then the largest segment of this marketplace is tier three, which is latency agnostic. These are guys that are still on fiber and they're price-point driven. They will take low latency if it is available at a price-point that is close to where they're at right now. So as a company, our charge is to take the same product set and drive it through all three of those latency tiers.  That is a challenge because you've got three price-points there, particularly when you're talking about bandwidth. The way that we've done that is with market data. Market data to us is the great equalizer. We charge the same for the biggest firms and the smallest brokers. It’s a way for people to get the same data at the same latency at a very palatable price-point. So while we started off as a carrier, I feel we've now evolved more into a market data provider. By the end of this year, transporting market data from one exchange to another will be our single largest revenue source. It has been a real sea change for us as an organization.

TR: How do you break into new markets for that kind of product?

MP: Clearly, we have to build the transport first, such as we recently did in Asia. The demand is initially for bandwidth, and that satisfies the tier one and tier two firms. Then, as that matures a little bit and there's what's called a reset in terms of what the latency is inter-exchange, we see people asking for market data. Transport begets market data in terms of demand and interest, because the folks that want the ultra-low latency are the ones that are pulling. The market makers and brokers of the world come along only when they have to.

TR: How is your Asian expansion going?

MP: Singapore is live and that network is up with customers on it. Shanghai is about to be live and we're really excited about Shanghai, because we were able to get some custom dark fiber done in China. That’s no small feat in a very opaque, difficult market to do business in if you're not an indigenous carrier with ‘China’ in front of your name. We think that that service is going to be pretty disruptive.  That market is burgeoning right now because of the INE, which is a subset of the Shanghai Futures Exchange, releasing a crude oil contract that can be traded by the external world. So now what you have is a natural arbitrage between crude at the CME and crude in Shanghai and that's driving interest in going to China, because it has been a closed market and you can't arbitrage in a closed market.

TR: Why choose to expand to these Asian financial markets rather than those in Europe?

MP: We chose Asia because we were able to acquire some assets that vaulted us into pole position for having the fastest known available networks between Tokyo and these other four regions. That was really attractive and opportunistic at the time when we were looking to expand. But also, maybe we didn't look as hard in Europe because it's a more competitive marketplace. It is congested in terms of assets and spectrum available and therefore expensive. One of the things that we didn't want is a "me, too" network. If you come in as the third or fourth entrant into a marketplace and you look like everyone else, all you have available to you is a pricing aspect. We don't ever think that that's adding value, whereas with Asia, we thought we could add a lot of value turning up some of the first hybrid networks. The other part that we liked is that our customers were expanding there. Most of our customers have been in Europe for decades, but lots still aren't even in Shanghai, and we liked the concept of going into this market together.

TR: What’s the biggest challenge you face in entering new markets?

MP: We live in a capex world, and we live in a green field build world. Many times, we go into places and there isn't a network there. There may be some fiber, but it may not be purpose-built. So, when we build something, we consider whether it will be the first of its kind, what it will cost, and what the take-up rate will be. But there are no pre-sales, and it very much is “Field of Dreams.” As much as you don't want to run a business on "build it, they will come," sometimes there is no choice but to do that. In the early days of building low-latency financial networks, there were some unscrupulous people out there that took all the advance money and then disappeared. We didn’t come over here because we thought there was a gold rush. I've been in the industry since 1999, so we have a bit of a pedigree and a track record.  But nevertheless, capex is the nature of the game, and the fact is that you do have to expend that before you get any revenue. We have been prescient and fortunate in terms of understanding the marketplace and building in the right areas, but it's certainly something we always have to be concerned about.

TR: What’s ahead of us in terms of low latency communications?

MP: People sometimes talk about the race to zero, and they say that it's over. I think that’s a little myopic or a little single-threaded, because the race to zero is actually multi-tiered. Sure, there have been a lot of networks built with wireless, which is usually the shortest distance between two points or pretty close to it, and you could say we are operating at the speed of light right now. That's not the point that I'm debating. But there are still two other pieces to this fabric that must be addressed holistically with latency: capacity and uptime.

 Suppose on the same transport network with a 200-microsecond one-way latency, one person has 100Mbps and the other has 1Gbps. The person who had a gig would be able to get more data there faster and effectively has a latency benefit. Capacity is the next battleground for "the race to zero." If you think about it as a highway where everybody can go 65, if you have 1 lane you can get only 1 car there at 65mph. If you have 5 lanes, you get 5 cars there in the same amount of time. How do we get fiber-like capacity over the wireless network? We've just recently released networks in New Jersey that are operating at 10Gbps. We think within a year we can move that backbone to 40Gbps. That’s really exciting because it truly is fiber-like capacity over the air.

When it comes to uptime on the fiber side, everybody talks about five nines. On the wireless side, the joke has always been, “well, it's wireless - we should just take nine fives, right?” These networks are up more than that, of course, but the fact is that they do go down. We've seen an expansion of these networks into every corner of the world, but it's still a challenge and people are still demanding higher and higher availability. I think if we can do that as a carrier, it gives us an ability to continue to push the product set down to the tier 3 customers that can't miss a packet. An institutional customer putting in orders for clients simply can't have the order not get to the market. At that point speed is irrelevant; it’s all about data delivery and data integrity.

TR: Thank you for talking with Telecom Ramblings!

Categories: Fiber Networks · Financials · Industry Spotlight · Low Latency

Join the Discussion!

1 Comment, Add Yours!


  • mhammett says:

    I think it’s important to note that the wireless networks being built by HFT guys aren’t the same kinds of networks built by fixed or mobile wireless networks.

    HFT wireless networks stretch the engineering to reduce the number of hops to get the best latency. All kinds of service quality mechanisms are dropped from the radios.

    Mobile and fixed wireless networks build for (typically) availability first, capacity second, latency third. A lot of these networks do have five nines of reliability.

Leave a Comment

You may Log In to post a comment, or fill in the form to post anonymously.





  • Ramblings’ Jobs

    Post a Job - Just $99/30days
  • Event Calendar