The future of stock markets is becoming less about speed and more about trading securities in different environments, particularly auctions. Several exchanges have introduced new auction models aimed at slowing the market down in different ways for different participants, including PDQ Enterprises, which is rolling out its new on-demand auction process for stocks. CODA Markets (Centralized On-Demand Auctions), the wholly-owned broker-dealer entity which was previously named PDQ ATS, will launch the the block trading facility called CODA Block in the next several weeks.
John Lothian News Editor-in-Chief Jim Kharouf spoke with Don Ross, who took over as CEO of PDQ Enterprises. He succeeds his father, D. Keith Ross, who moved to executive chairman. Ross, who previously served as PDQ’s chief strategy officer and spent 11 years at GETCO (now KCG Holdings) spoke about what is wrong with speed trading in today’s equity markets, why institutional participants need a new platform and how CODA might just help bring all the key participants together.
Q: For those who don’t know you, tell us a bit about yourself.
A: I’ve grown up in electronic markets. My first job out of high school in 2001 was as a summer intern for GETCO. when there were about 10 guys in the company. After a summer of making spreadsheets and getting coffee, I learned how to use the first version of their proprietary trading software and they offered me a job to be the third algorithmic trader. I deferred my college education for a year and devoted myself full time. Then I ended up getting into Northwestern University, which was conveniently co-located where the trading office was in Chicago. For three years, while I was earning an organizational change degree, I worked two or three days a week on the testing, documentation and building training programs for traders on how to use our ever-evolving software.
I traded full-time during the summers and when I graduated in 2005, I came back to GETCO to trade full-time on the eurodollar desk for a while. In 2007, I built out our commodities trading desk when those markets were going electronic. Then I turned back to equities where I started and eventually became the co-leader and then sole leader of our American equities and technology group.
I spent 11 years at GETCO and that was an amazing time. I learned a lot about how hard this work is, as well as what happens in companies as they grow and the complexity increases, and about the management systems.
Along the way, GETCO made an investment in PDQ when it was run by Chris Keith – around 2003. From that moment, the value of this options-based market structure was clear. There is a lot of benefit to the risk profile of both sides of the trade. It’s just fundamentally a more efficient way to trade. But it definitely diminishes some of the incentives of the speed arms race that GETCO was actually winning for a very long time. Even though GETCO was an early investor, PDQ didn’t actually play to our strengths. GETCO never got involved with the trading on it.
But flash forward a couple years ago, after I moved on from GETCO, PDQ was growing into a real business with real market share, profits and a platform to have a real impact on changing the way people trade. So for me, it was the combination of what I know about how trading works at a market structure level, and in terms of how to lead the product development and research side.
We’re still a small company, but we’ve been growing, and we’ve had enough success that we have had to implement a little more of that management process. After seeing some very good things and not so good things about how we approached that at Getco, where we went from 10 to 400 people, I feel like it is the perfect opportunity to apply my experience and make a difference.
Q: How so?
A: In my mind, the massive failure in today’s market infrastructure is in providing the opportunity for an institutional block trader to competitively negotiate a block outside the spread, with a diversity of market participants. There simply does not exist a way for an institutional participant to do that [in a market] which is dominated by a continuous order book. You can’t actually explore the latent liquidity outside the spread without tipping your hand that you are someone interested in trading the spread. So delivering a market structure that does that is something I care deeply about, and something I can have an effective impact on the market.
Q: What is it about that area that provides opportunity? Is it the technology side of the business that enables that, or something else?
A: There are a few different pieces to that puzzle. First, you have to create a trading system that allows the right dimensions of competition. That has been part of PDQ’s DNA forever. The idea of pausing an order and soliciting responses from multiple participants is the idea that Chris Keith patented. But applying it to the block trading problem required quite a bit of design and implementation work to make it a system that works for high touch traders and low latency traders.
It’s getting an auction that’s long enough that buyside guys can put in with IOIs (indication of interest) and algorithmic guys can reply but their order isn’t held up so long that they can’t handle the risk. That is a challenging design issue. None of that is simple.
There is another piece to it, which involves our ALISA research platform. We hired two of my former GETCO colleagues to help build it over the last year. That’s building in some safeguards about the terms you are going to be applying to your block liquidity. When you start trading real big size the stakes get pretty high, pretty quick.
The fast market structure allows participants to express their most aggressive order – biggest size and most aggressive price – and trust that the competitive forces are going to get them the best fill. In order to engender that confidence, we thought that it was important to have some data driven safeguards. The first version of that, which we have built into CODA Block, we call the liquidity protection rule. We use ALISA to look at 20 days of trades for every single stock and to calibrate individually for each stock an equation that characterizes price variance as a function of its notional turnover, which is just another way of saying we develop a curve that says, if you’re going to trade 1 million shares of Apple, what is a reasonable distance to move the price outside the spread? Or at what point would you say that is an unreasonable distance at the price given the size?
On our website, anybody can click on the ALISA portal, sign up and take a look at what we call our liquidity protection rule calculator. That gives transparency in how that pool will operate inside CODA Block. It’s updated overnight and with fresh data every morning with the latest day’s trading data on every single stock.
Q: I hear about this market structure problem a lot, that in this key segment of the marketplace – the institutional side or even the market making side – participants have been displaced. Does this provide a place for them?
A: We believe the best market is the most inclusive market. It’s the diversity of participants which makes for great trades. If I have an apple and you have an orange, and we both prefer the other person’s fruit and we trade, we’re all better off.
If we’re all buyside guys with the same kind of portfolio, we’re more likely to be on the same side of things, which means we don’t trade very often. But when you get people who are trading at very different time horizons, there is a huge opportunity for them to trade with each other. And both should be very happy about the terms.
But there are a lot of things about the market structure that make it hard for those guys to meet outside of the 1 round lot at the NBBO context. We think this market structure makes it easier for everyone to participate. Certainly on the provider side, you get an opportunity to participate with every incoming liquidity-seeking order without having to expose yourself to the open order risk and adverse selection on every other continuous limit order book market. That is a huge cost for the providers, the risk they have to take, to have a chance to trade with incoming orders. And it is a huge benefit not to be exposed to adverse selection where you try to cancel an order only to find you’ve already been filled on it. That doesn’t happen on our system.
Q: So how does the market participant base look today and how would you like it to look?
A: We have over 150 subscribers, substantially more sellside. We are obviously trying to get a lot more business from the institutional buyside community. But we expect a lot of that to come through the brokers and they are valuable partners in getting that out. As far as the mix, we haven’t had a block liquidity solution before now and I hope it becomes bigger than our existing business.
Q: When you look at the overall securities market today, it’s extremely fragmented and competitive. What does it take to be successful?
A: Right now, we execute and route over 1 percent of the average daily volume. That is a level of success that we are quite proud of and that gives us a chance to have much bigger impact. It’s not unreasonable to think of institutional block trading as a niche. I would argue that strictly only applies to the liquidnet (or luminex) paradigm of natural to natural block trading. There is obviously a low ceiling, with the amount of volume that is traded that way. It’s limited by the naturals who are on the opposite side of the trade at the same time. When you are talking about a market that brings in every type of participant through a competitive auction process to assemble block liquidity from a bunch of fragmented parts, there is absolutely no limit to the amount of volume that can trade that way, essentially because it is a more efficient way to do it and it lowers the cost liquidity and increases demand.
If we got another one percent and were trading another 50 million to 100 million block shares a day, that would be successful. But that is not the full potential of what we think our marketplace stands for.
I completely agree with your description of this fragmentation of liquidity. It is endemic of an incorrect paradigm about how to use modern technology to create an efficient market. Instead of liquidity seekers having to hunt across the whole world to find the liquidity they need, we’re trying to create a market where the liquidity comes to you. And between our block option and our micro option, and our routing capabilities, I think we’re developing a one-stop shop solution. That is the level of my ambition.
It’s fair to put participants into two categories – buyside and sellside. But to me it is this: are you in the market because you have a position you are trying to achieve? Or are you in the market because you are willing to take a position for the right cost? If you are in the market for your position and you go to PDQ, we want to deliver the best fill you are going to get. And if you are a pro, you want to come here because we give you access to the people willing to give up edge to trade without having to win the speed arms race. That is becoming an obscene barrier to entry for people who want to get into that game.
This isn’t to diminish the need for speed for market makers, and there is nothing evil about those guys wanting to be co-located and build an FPGA (Field Programmable Gate Array) hardware accelerated network system. They need those nanoseconds because they are taking the open order risk, and it really hurts when cancels are slow and you get run over. But the result has been improvements in latency which create zero value for the investors who have positions. Those guys don’t give a crap about a nanosecond. So this is a market that prioritizes liquidity over speed, designed in a way where you have the ability to trade with the guy with the best price and the best size, and not just the guys who get there a nano- or a millisecond sooner. I think that pushes the market into an entirely new level of efficiency that compares to the step change in efficiency we saw with decimalization and electronification of markets 15 years ago. That was a massive improvement. Unfortunately, the one or two magnitudes of change are accruing to retail guys. And we have a plan to bring that level of efficiency to every participant.
Q: Other competitors are trying to address the speed issue as well – IEX and Chicago Stock Exchange, for example. How do you differentiate yourself from them?
A: The IEX comparison is very simple. The extent of the similarity ends at, we both slow down an order. But when IEX slows down an order, they do absolutely nothing to change the fundamental perverse incentives of the speed arms race. They still process orders in the order in which they arrive, and they process each one discretely.
Eric Budish at the University of Chicago did a white paper explaining that every market operates a continuous limit order book that processes orders serially, in the order in which they arrive, creating a situation where, even if every market participant observes the data at the same time and responds with equally fast lines and hardware, invariably one of those orders gets to the exchange before the other orders and that order wins. So it’s the “if you ain’t first, you’re last” paradigm. That’s the order that picks off the slow one. Or that’s the order that gets cancelled and doesn’t get picked off by the other guy. The fact that those arbitrages then can be derived from symmetrically observed information is the source of the major deadweight loss of our existing market infrastructure.
IEX maybe helps with the latency arbitrage between fixed speed and the direct feed. But even if you solve that perfectly, you’re still going to do nothing for the fact that orders are processed in the order that they arrive and if you are a nanosecond ahead, you get the whole trade and the guy behind you gets nothing.
Our market is fundamentally different from that, in that when we pause an order, we send out an IOI and assemble competitive responses into an auction book. And it’s not until we get all the responses back that we prioritize them based on price and size. So our market prioritizes liquidity over speed. IEX thinks they are trying to do that but they don’t come close at all.
Q: How long does the CODA Block auction run?
A: CODA Block runs for 30 seconds. We’re guessing that is a good amount of time. It has to be long enough that human traders can actually receive an IOI and respond to others. We’re trying to create a way for some participants, such as natural block liquidity consumers, to access the market makers who are the electronic suppliers and the most efficient suppliers of liquidity. But I don’t want you to pay those guys for liquidity if you can find a natural participant who is willing to trade at the midpoint. The first 29.9 seconds are all about assembling other high-touch interest liquidity. So we send out a symbol-only IOI to most of the high-touch traders as well as those that use algorithms.
We have one version of our micro-auction that can be between 5- and 20-milliseconds. And we have CODA One – that is a one second auction for conditional orders. But all of those involve a symbol-only IOI. You don’t get the size of the initiating order. You don’t get the price they are willing to pay. You don’t get the size they are looking for. All you know, for example, is that we’re trying to trade IBM. So we spend 29.9 seconds assembling all the interest in IBM that we can get and then we ping the algorithmic liquidity providers for their market and we put that into an auction book. We figure out at what price the greatest number of shares will trade.
Q: IEX and CHX are looking at 350 microsecond periods, which is a very different time period than yours.
A: I’m not the CHX expert but I believe they can end up holding a market maker order as much as 500 milliseconds, which I think is too long. It is an auction. In terms of diverting the perverse speed incentives, it gets a much better grade than IEX. But we’re building a market infrastructure that allows high-touch traders as well as low latency traders to participate in a way that doesn’t force either to give up their best and most aggressive prices. The CHX Snap isn’t long enough to get other high touch traders involved apart from the guy who initiated it. And I think it’s too long to get the participation from the algorithmic guys.
Q: What’s been the feedback on CODA?
A: The feedback has been overwhelmingly positive for CODA Block. We’ve had a couple guys focused on the institutional world for a while and now more of our sales team is getting involved in that product. I keep hearing that this thing sells itself.
There is a lot of work needed to convert that enthusiasm into orders and participation. Chris Keith had this fundamental idea 15 years ago and it’s taken this long for the market to be ready for it. But I believe we are ready for it now.
Q: As you look ahead, what is the potential for this space?
A: I think we can fundamentally change the way people trade and re-aggregate liquidity. What that means for our market share? I don’t think there is a ceiling.
At 5 percent market share, the fair access rule kicks in, but we do provide fair access to our markets to everybody and don’t have an order book to display. So I don’t anticipate that being an issue. I want to get to five percent not just on a couple of symbols but on a lot of active symbols, since we’re putting up all the big trades. And then I want to tell that story and make it clear that there is nothing for anyone to worry about trading on our market.
We report trades in a timely fashion and we’re going to produce oodles of data proving the excellent quality of execution that people get on our platform. When we stop wasting money on the speed arms race and can create lower cost liquidity solutions for the big boys, who are the real liquidity consumers, I expect to see more volume.