Doug Rivelli, co-chief executive officer at Pragma Securities, spoke with Traders Magazine recently about several issues affecting equity trading. Among the topics discussed were trends in algorithmic trading, future regulation, high-frequency trading and "quote stuffing."
Traders Magazine: What is your biggest concern, or what has your attention right now in the electronic trading marketplace?
Doug Rivelli: The biggest concern in general has been declining volumes in the marketplace. We obviously see that as the biggest threat to all firms, whether they are electronic or non-electronic. That said, specifically in the electronic space our biggest concern is the regulatory environment and where regulators are going to go in terms of regulating dark pools or, say, high-frequency trading.
TM: What issues have regulators raised, such as rule proposals or concepts, like Trade-At, that you are looking at?
Rivelli: I wouldn’t say it’s been a specific regulation that has been discussed–we’re tracking all the proposals closely, as well as the dialog around the concept release and high frequency trading. One thing that does concern us is what seems to be a strong drive to regulate that’s shaping up. It’s notoriously hard to tinker with the markets without having unintended consequences-look at decimalization and Reg NMS for example-and there seems to be a particularly strong pressure and appetite to regulate right now. There’s a lot of concern and insecurity around high frequency trading in particular, but many of the ideas floating around seem arbitrary, and may have more unintended consequences than intended ones.
TM: How so?
Rivelli: As an example, we’re waiting to see specific proposals, but there’s a lot of talk about gaming. If you regulate any firm’s ability to, let’s say, identify supply and demand imbalances in the marketplace and take advantage of them, you’re potentially eliminating the entire reason to have a trader. To regulate something just because it’s turned from a human activity to an electronic activity tends to concern me a lot. As another example, there’s talk about changing rebate structure. Eliminating liquidity rebates would definitely make things harder for high-frequency market makers, but would have significant effects on exchanges and on available liquidity, and it’s hard to know what the secondary effects would be. Ideally, solutions to some of these problems and the concerns in the marketplace should be addressed by the commercial market.
TM: What do you make of the SEC’s rule proposal to kill unfiltered sponsored access?
Rivelli: I think that the problem with sponsored access is a lot of orders are bypassing robust risk control systems and I think that poses a big danger to the financial system. Every order that comes into our infrastructure is checked for size of order, total notional dollar value they’re able to trade per day, previously placed orders before they are sent to an exchange, credit limits, etc. Those types of checks are critical to avoiding not only a "fat finger" error, but in keeping a client or a firm that is doing a lot of trading from defaulting in a way that will reverberate through the system. Risk controls are very, very important in the electronic space.
TM: How, if at all, have high-frequency traders affected Pragma’s business?
Rivelli: Generally they are affecting our business by requiring us to maintain a very rigorous ongoing research and development process, though we should really clarify what we mean by "high-frequency trader" because there are really several distinct categories of player we could talk about. We’ve put a lot of effort into understanding how market-makers work, for example; how to best interact with their liquidity for our clients; how to benefit from it and what challenges it poses. Another category of HFTs is what we call gamers. They have driven a lot of our research into anti-gaming and execution quality monitoring. As they devise new strategies to sniff out and exploit institutional orders, we need to be able to quickly develop countermeasures for our clients against those strategies.
TM: Should there be quoting obligations for those acting like market makers?
Rivelli: First of all, again we should start by asking: what is high-frequency trading? Presumably only players that behave like market makers, though how the group is to be defined is unclear–unlike true market makers in the past, they’re not getting regulatory benefits that they have to register for so they might not identify themselves. This is where we wonder about unintended consequences. Who will the regulations apply to? What will be the effect on competition? And will it all actually prevent another flash crash? None of this is clear to us at this point.
TM: There is increased talk lately about so-called "quote stuffing," and Nanex’s research report talking about it. How much of an issue is quote stuffing? How do algos cope with this type of practice?
Rivelli: Quotes flashed into the market and cancelled have been part of the trading landscape for some time. The difference today is the speed and volume of the messaging. These behaviors have a very limited effect on the workings of our algorithmic strategies. We look at many factors when making microtrading decisions, including quote size and duration, spreads, hidden order detection, overall stock liquidity, etc. So, even a large number of quotes with such as short lifespan will not adversely influence our microtrading decisions. Processing all the message data from these quotes is not an issue for us at this point.
TM: What is the latest trend in algorithmic trading? Is it unique algorithms designed for specific customers or staying with "one-size-fits-all" offerings?
Rivelli: There is a definite trend in our business, at least on the agency side, toward the concept of bespoke trade process analysis for clients, and based on that, developing customized algos designed to meet the specific needs of a trader, portfolio manager or trading strategy. We spend a lot of time with our clients doing just this. This kind of deep analysis can reveal real insights to improve trading performance, and often we can embed those insights into a customized algorithm. We often hear traders talk about a disconnect between the stock selection process and the execution process. This process of analysis and creating bespoke algos can help bridge that gap.
TM: Explain that, if you could.
Rivelli: An example would be where a PM has an optimization process and comes up with a trade list that has very specific risk characteristics-whether it is a buy/sell imbalance, some sector over-or underweightings or some specific tracking error they’re trying to get to. Often what can happen during the execution process is if a trader makes a bet, or if they are pulling out individual names and trading them in a different way than the rest of the portfolio, then in the middle of that trade the risk characteristics might look a lot different than what the portfolio manager had intended through the optimization process. By building bespoke models, we can control that kind of risk profile throughout the course of the trade so that at every point in time the risk characteristics are going to be in line with what the portfolio manager is trying to achieve. That’s just one example of the benefits of building these bespoke algorithms.
TM: The buyside has expressed a desire for more control of where its orders are being routed and executed. How do your algorithms help them achieve the goal of transparency?
Rivelli: We’ve been very loud advocates of transparency in the way algorithms work, in terms of how and where orders get routed. Is internalizing an order really the best thing for the client? Does an algorithm route an order to an internal engine and then maybe route it to a venue where brokers don’t have to pay an execution fee, and only after that do they route it to an open market venue where they pay a fee to execute? We believe clients should know all details about where orders are being routed and why.
TM: You have stated previously that mid-size sellside firms were a target for Pragma? Is that still the case?
Rivelli: Yes it is. I don’t think a lot of firms realize the amount of time and effort and resources that go into building and maintaining a robust suite of algorithmic products-it’s something that even very large firms find challenging, and Pragma can act as a real partner to firms who might be luke-warm on sending order flow to a larger competitor. Pragma also provides algorithms for exchanges. Our algorithms, for example, are on the floor of the New York Stock Exchange and available in hand-held devices to floor brokers.
TM: Do you have plans for further expansion?
Rivelli: For the broker-dealers, yes. Our broker-dealer offering is designed to allow clients to have total control of where their orders are being routed by either using their own MPIDs or they can use the Pragma infrastructure if they want more of a "turn-key" solution. We also continue to work very closely with the NYSE to work to build innovative products for the members of their floor community.
TM: What about expanding business outside the U.S. markets? Are there any new business lines Pragma is exploring, such as research or branching into new asset classes?
Rivelli: We are in the process of developing non-U.S. equity algorithms. We are also looking at expanding into other asset classes. In 2011, we expect to have growth both outside of the equity space and in global equities.
TM: Are you hearing about (or seeing) more blocks trading in dark pools?
Rivelli: I think that volumes over the summer and in general have been relatively light. But what I will say is that clients do express a desire to trade blocks. I’ve heard that at some venues there is some concern about counterparties and how the information that goes out prior to a block trade is controlled, but at the end of the day, our clients are telling us they would like a resurgence in their ability to trade a block. I would suspect if volume builds throughout the fourth quarter, you’d see an increase in the number of blocks printing on the tape.
TM: Are you making any tweaks to OnePipe based on client new demand for algo products?
Rivelli: OnePipe is our cornerstone product in aggregating liquidity in dark pools, and we’re in the middle of a major push to fully integrate that liquidity-aggregation capability into our whole suite of algorithms. Our clients, and particularly block- and single-stock traders, want to source liquidity wherever it’s available, lit or dark, in an intelligent way.
