The Gateway to Algorithmic and Automated Trading

In the worx: What Manoj Narang thinks about regulation, Europe, volatility and a lot more

Published in Automated Trader Magazine Issue 29 Q2 2013

Manoj Narang wears a lot of hats. He's a technologist who built a low-latency platform and developed a monitoring system for the US SEC. He's an executive who runs a hedge fund and an HFT group. And he's an outspoken champion of data-driven decision-making. So why is he promoting ideas that could hurt his business?

Manoj Narang wears a lot of hats. He's a technologist who built a low-latency platform and developed a monitoring system for the US SEC. He's an executive who runs a hedge fund and an HFT group. And he's an outspoken champion of data-driven decision-making.

In this interview with Automated Trader Editor Adam Cox, the CEO of Tradeworx talks about why he'd be happy to see some market changes enacted, even if that meant hurting his own business.

Adam: Your activities are now a lot more extensive than they were when you first formed the company, so what is the focus at the moment?

Manoj: In the beginning our focus was similar, conceptually, to what it is now, in the sense that we wanted it to be a democratising force when it came to availability and applicability of financial technology. But our principal way of making that a reality, when we first started the company, was to focus on the retail investment segment. And that's because the retail segment was actually where the locus of innovation was back then, because of all the changes in the brokerage industry, namely the advent of ECNs, or ATSs as they're known, the advent of online brokerage and of the internet as a data dissemination medium. All of those were very, very democratising forces in their own right and it opened up an opportunity that we saw as basically bringing analytical decision support tools directly to retail investors to empower them to make more efficient and more optimal decisions that correctly reflected their beliefs, opinions and preferences, and also, behind the scenes, leverage a lot of the sophisticated machinery of modern portfolio theory and metrics that they may not have that much familiarity with, in a way that was rather transparent.

The closest analogy is Google. The Google search box is simple enough that billions of people can use it. Really, the only requirement to use Google is just basic literacy. So if you can formulate a search query, behind the scenes some very complicated algorithms and mathematics are brought to bear to bring an optimal answer. But that's all beside the point as far as the user is concerned. They just care about the applicability of the answer. We had the same perspective.

Because of adverse operating conditions related to the bubble and the bear market of 2000-2001, we changed our business model circa 2001 to focus on institutional priorities. One thing we did was start our own hedge fund, which continues to have a superlative track record. Our mandate was always use completely home-grown technology, both on the trading side and the research side, and monitoring side. By doing that we've also created a very powerful base of intellectual property that can be re-appropriated by making it commercially available.

We started another business, namely the high-frequency trading business, in response to adversity in the hedge fund business around the time of the financial crisis. So we now have two different trading businesses. They both have different domains of intellectual property that give them their edge and so, wherever possible, we seek to commercialise parts of our IT that we deem to have wide applicability without compromising our edge.

Our high-frequency trading strategies are all dependent on an ultra-low latency trading infrastructure that we built from scratch, which is among the most powerful and complete in the industry. At the same time that we rolled out our top trading strategies, we also rolled out this platform for wide commercial availability and became clients of it ourselves. All in, about 5% of the average daily volumes of the US equity market is now going through that platform. We are one of the users of it, but we are not the largest.

Adam: So that 5% refers to all the users, not just yourselves?

Manoj: Correct.

Manoj Narang, Tradeworx

Manoj Narang, CEO, Tradeworx

Fear, paranoia and ignorance

Adam: You've been described as one of the most outspoken champions for HFT. Is there a reason for your passion about this subject, apart from the fact that it's your livelihood?

Manoj: There are a whole lot of reasons why I've been very active in terms of reaching out to policy makers of all stripes and the press and the public. Principally it's because I feel there's a tremendous amount of misinformation and my interest is in a better marketplace. I think that when there's fear and paranoia and ignorance and ill-founded statements and basically lack of understanding out there - and that's driving policy making - I think that's very dangerous to the markets. So it really has nothing to do with my livelihood per se. I'm sure that some of the reforms that I've been advocating would be detrimental to my current high frequency trading business, but our interest is in a fair level playing field market and in a properly functioning market. I think that any sort of regulations, rules, demarcations to the current market structure that facilitate trader fairness and transparency - we support those whether they're beneficial to us in particular or not. That's been our motivation all along.

We are definitely concerned about the prospect for policies that are undertaken for the wrong reasons. Those reasons could be cynical reasons, having to do with grandstanding and so forth, sort of what you've seen in Europe for the past couple of years. Basically, what we don't like is that, rightly or wrongly, the public view of the financial industry has really plummeted since the financial crisis. I think much of that is well-deserved but the problem with that is there's been very little differentiation between good actors and bad actors in the industry. And what I don't like - and what I tend to take personally - is being lumped in with people who had something to do with the financial crisis. Nobody ever heard of high-frequency trading, nobody ever cared about high-frequency trading before the financial crisis, and because the financial crisis happened to coincide with when people first heard of high-frequency trading, I think those things got conflated, even though they have nothing to do with each other whatsoever.

The other reason is that the public's understanding of the financial world conceptually begins and ends with Wall Street. And Wall Street is more or less synonymous with the New York Stock Exchange. And the New York Stock Exchange is more or less synonymous with the stock market. So the point is that the stock market had little if anything to do with the financial crisis. The financial crisis was a crisis of illiquidity in the over the counter derivatives markets, in particularly the mortgage derivatives market and other credit derivatives markets. And also the fixed-income markets in general, particularly in Europe as they relate to sovereign debt.

Manoj Narang, CEO, & Di Liu, quantitative strategist, Tradeworx

Manoj Narang, CEO, & Di Liu, quantitative strategist, Tradeworx

Those things have very little to do with electronic trading. Electronic trading is actually the cure for what ails those markets and that's why the Dodd Frank regulation that was spawned by the financial crisis calls for so much more electronic trading of those kinds of securities. The reason the equities market withstood the stresses of the financial crisis was because of the fact that it is a transparent, fully-electronic market.

We definitely are in favour of market regulation to make markets more robust and more stable and which fix some of the obvious flaws, but on balance our belief is that the US equities market is one of the fairest and most smooth-functioning markets in the world. Yeah, it's always good to tweak and make things better where it's obvious how to do that, but going back to the two-tier system of before, with wider trading spreads and larger transactions cost for investors, is not the way.Now, the equities market has had hiccups, for sure, since the financial crisis. There was of course the Flash Crash, et cetera. One has to question how much you want to do about that. A lot of the reason why those things have happened is because of the rapid pace of innovation, both in regulation and technology. On balance those are good things. And when things like the Flash Crash reveal themselves, there are natural fixes, most of which have already been enacted such as coordinated circuit-breakers between the markets. There are other regulations that take the market to a better place also. But what we don't like is an extreme reaction in the opposite direction, where we want to go back to a market with privileged insiders, to start walking down the slippery slope back to yesteryear. The whole reason we had all these regulations in the past 20 years was because those were not good times. Investors had significantly higher transaction costs, insiders were much more profitable and privileged than they are now. Now it's pretty much a level playing field market. People who are involved in market making or high-frequency trading don't have any sort of advantage that anybody else can't have. Whereas in the past, your status as a market maker really gave you all kinds of unfair advantages.

Adam: Another factor in the crisis of course was the ethical dimension, with different market participants having incentives to behave a certain way - and the electronification of markets also has nothing to do with that either.

Manoj: The derivatives market every five or six years has had to stop and blow up. We saw it with Long-Term Capital, we saw it just recently with the financial crisis. And the Long-Term Capital thing of course was coincident with all the issues with Russian debt and so forth. But the point is that every so often, these markets imploded on themselves. You just don't see that with the equities market. So we think all of the attention focused on reforming the equities market is misplaced. Really, what ought to be focused on, if you want to ensure financial stability, is the derivatives markets and make them more electronic, not less.

Adam: That's what's happening now under Dodd-Frank, wouldn't you say?

Manoj: Yeah, under Dodd-Frank that's been happening. But we think that the attention of press and policy makers on balance has been largely misdirected to fixing things that are not broken.

'Volatility is never dead'

Adam: Staying with the HFT theme, one view is that the 'low hanging fruit' is pretty much shrinking as volumes have ebbed. Is that what you're seeing?

Manoj: There's really no mystery here. I've been trying to explain this to people for years now. When HFT was at its peak in 2009, everyone thought that it was going to expand and - quote-unquote - take over the world. And everyone was very worried about that. That's not at all what the reality of the situation is. The reality of the situation is that high-frequency trading is a natural long-term participant in the market in the sense that it's always been there. It used to be called something else, but the trading style was just as high frequency as it is now in the sense that there were always liquidity providers. They went by names like 'specialists' and 'market makers' and 'floor traders' and so forth, but their objective was always the same. They extracted a small spread out of the market in exchange for supplying orders and volume that allow long-term investors to more effectively make their bets. These were not people who were taking long views on stocks; they were looking to take the spread out of the market or to scalp the spread between the correlated instruments. In fact, you want to liquidate your inventory as quickly as possible in those strategies because it's adversely selected, so you're not trading based on any information about the stock, and so you get run over by large investors unless you manage your inventory carefully.

All the hype and paranoia about high-frequency trading is completely misplaced. All high-frequency trading today is an automation and electronification of that long-standing strategy which has always served a very useful purpose.

Now in terms of the amount of high-frequency trading that's going on, it hasn't really changed. It's always been about 50% of the market. And the reason that it's always about 50% is that, on balance, high frequency traders don't trade against each other. They tend to run very similar strategies. They tend, for example, to be passive traders rather than active traders. In other words, people who are posting limit orders onto the order books of the exchanges, rather than removing limit orders from the order book. Because of that, there's a natural upper bound of 50% that their volume can add up to. If there's a group of market participants that don't ever trade against each other, then that group cannot possibly mathematically do more than 50% of the market volume.

So when market values ebb and fall, high frequency participation ebbs and falls with it. It's completely commensurate with the overall level of volume in the market. The reason for the compression of profitability is two things. Number one: volumes have gone down. And let's be very clear about something: high-frequency traders cannot control how much volume they do. They're principally passive participants in the market, and what that means is they don't control when they get into a trade. The active participant, who's typically a buy side investor, is who controls when there's a trade. So we can wish there to be more volume all we want, but unless there's somebody who's there transacting against our quotes, there will be no life.

Just as it is the case for firms in the oil industry, when there's less demand for energy, oil companies become less profitable, oil prices fall, et cetera. When oil prices skyrocket, all of a sudden there's room in the market for new participants. That's a reflection of the fact that the supply of that commodity is relatively inelastic, but the demand for it is highly variable. The same thing goes on in the equities market. The supply of liquidity is roughly constant because the players who are providing liquidity have a finite base of capital and the demand for liquidity is varying all the time. The higher the demand for liquidity gets, the higher volatility gets because of this discrepancy between the elasticity of supply and demand. So when investors are suddenly demanding 10 times as much liquidity as they have typically, there's not suddenly 10 times as much capital there to satisfy their demands. And so prices move more. That's what we saw in 2008-2009. These tend to be very profitable environments for providers of liquidity because it's reflective of the situation where liquidity is at a premium. When you're selling a commodity that's precious and in demand - in this case it's liquidity but it's immaterial, it could be energy or corn, it doesn't matter - you tend to make more money than when your commodity's oversupplied.

The profitability crunch is because volumes are down and at the same time volatility is down. The VIX is near historic lows. But that's not a permanent thing. What people have to understand is that firms in our business that go for the long run have a long view. When volumes and volatilities do return to the market - you know, volatility is never dead, it comes and goes and right now volatility is low - the business will get more profitable.

Adam: There are a few factors people have talked about with regards to volume.

Manoj: The thing is that volume hasn't really gone down. If you look at dollar volumes, they have gone down very marginally since 2008, to roughly constant, year on year. Share volume has gone down because prices have gone way up.

Unfortunately for high-frequency traders, it's a transactional business. The more share volume there is, the more profits are made. Dollar volume doesn't really help. But from the perspective of the health of the economy or the health of the stock market, it's functioning just fine.

Different behaviour

Adam: Within that environment, as an HFT firm, are there specific techniques which will allow some firms to thrive more in this environment until volatility starts to come back?

Manoj: I think profits are down for everybody regardless of what their expertise is. So what you're seeing right now is a lot of cost-cutting, a lot of mergers and acquisitions that are intended to exploit economies of scale. You know, if you can double your volumes through acquisitions and cut part of your costs, then you end up inflating your profits, since you're doing double volumes for double the revenue with less than double the cost. That's a lot of what you're seeing right now.

In terms of how that bears on technology, you see firms behave very, very differently than they behaved in 2008-2009. In those years, there was a tendency to overinvest in technology and everyone was basically demanding their own proprietary R&D budget, re-inventing the same wheel as everyone else.

Now what you're seeing is that firms that specialise in technology are basically able to sell technology to lots of people because the larger firms are no longer making those kinds of aggressive investments in R&D, as their profits testify. So this is definitely something that we saw coming and that's one of the reasons why we decided to commercialise our technology.

Adam: One of the findings from our own survey is that there may be a greater focus on latency involving decision-making and latency in terms of capturing trades and factoring them in a system. What are your thoughts on that?

Manoj: There are many ways to conceptualise the sources of latency and put them into abstract buckets like that.

In general, what is true is that latencies can come down in all areas, and as they get closer to zero, it costs more and more. The lower latency you get, the more expensive it is to drive it down further.

You can't get less latency than zero, therefore the lower latency gets, the less you can reduce it. That sets you up for strongly diminishing returns. It costs on margin more and more to get a smaller and smaller reduction in latency. Concomitantly with that, as you get closer and closer to zero, the less deterministic those improvements in latency are. In other words, you can reduce the mean latency faster than you can reduce the variance of the latency. My point is just that at some point it stops becoming economical to obsessively focus on latency. There are other things in life that are far more important to success or failure in investment strategy than latency.

Adam: What are some of those other areas?

Manoj: Your alpha signals for one thing.

Adam: Are you talking here about the hedge fund business rather than HFT?

Manoj: No, HFTs do the same thing as quant hedge funds. They just do it for a different domain of correlations. But ultimately all these quantitative businesses are all about modelling correlations. It's just different kinds of correlations. HFTs tend to dabble in correlations that are much more stable because they are of a structural nature: for example, the relationship between an index and its underlying stocks or the relationship between an ETF and its underlying index or the relationship between a futures contract and the underlying index, or what have you.

These are the bread and butter strategies of HFTs and they're exploiting correlations that are very, very stable over time and therefore easy to model. Because they're easy to model, there's no particular skill set required to model them effectively, so it becomes a game of who can access the opportunity the fastest. But there's a different domain of correlation trading called statistical arbitrage, where the correlations are much more difficult to model because the historical correlation is not stationary. It's constantly time-varying and so the spreads between correlated stocks can grow rather wide because you can never be sure how much of the divergence is temporary and how much is permanent.

Those spreads can get very large and therefore you can accommodate very large bets, you can deploy lots of capital on those bets and they're not particularly time sensitive. Those are two extreme points, but in between those two extreme points, there's a very healthy spectrum of opportunities. My point is that high frequency traders and statistical arbitrage traders are basically the same animal. They just dabble in different correlations, and as a result, different skill sets become important. But that's just a conceptual difference that exists at the extreme ends of the spectrum. Increasingly they're starting to impinge on each other's turf.

Adam: That's interesting that they're starting to go after each other's turf.

Manoj: Well, our firm does both things. We are quite good at the ultra-low latency end of the spectrum. We also have a very established statistical arbitrage business that's going on its 10th year now and we definitely see that techniques of both businesses are applicable to the other side. We think that the real room for growth in these fluctuating businesses is in the centre.

Data-driven decisions

Adam: You've publicly said you would not trade in Europe given the regulatory climate. Do you feel that has particularly limited you and held you back from markets you'd like to pursue? And what would it take to change that?

Manoj: There's no doubt we'd like to trade in Europe. Those markets are not nearly as liquid as the US stock exchange. And they're destined to get even less liquid due to all these transaction taxes.

Stocks that are above a certain size in market capitalisation are taxed, but stocks that are below that are not. And there's evidence that the liquidity has grown by something like 16% in the less liquid ones, and has declined in the large caps because of the tax.

In the United States, the situation would be very, very problematic because there's a far greater liquidity premium built into the value of US stocks than there is in European stocks. We're a far more liquid market. And if you get rid of that liquidity premium, you're going to see a catastrophic fall in asset values in the US, because of the fact that there is such a large liquidity premium. The asset management industry here, all the hedge funds, all the mutual funds that depend on a liquid, vibrant stock market, those guys are not going to be able to support nearly the assets that they have currently under management if the liquidity margin shrinks by 30 or 40%. You're going to see massive, massive outflows from those types of asset managers. It's going to create this gigantic, catastrophic wave of volatility, in my view, if something like that were ever put in place here. I don't think it will be. I think that the policy makers and the regulators here are far smarter. There are some less smart, if you will, members of Congress who don't really understand how markets work, but on balance, our regulators tend to have a much more measured approach to decision making and policy making than their counterparts in Europe.

In Europe, it tends to be 'act first, think later'. Whereas over here, you have the chairman of the SEC promulgating this notion … she's really advancing or advocating the concept of data-driven decision making, and how the SEC's recent data collection and data analysis initiatives are really going to form the underpinning of future policy. That to me is a very salutary thing - that's a cause for optimism and a cause for a return of investor confidence in the market.

Adam: That leads to the next question. You're a somewhat unusual firm in that you're also a supplier to the SEC, in terms of the market oversight technology you developed for the regulator.

Manoj: Usually firms that are involved in trading have a somewhat adversarial relationship with the regulators. They're worried about penalties, they're worried about secrecy, all sorts of things like that. We're mostly worried about a fair and efficient market.

Adam: How did it come about?

Manoj: It's pretty straightforward. The SEC launched a very widely circulated and widely read publication in 2010 called the Concept Release on Equity Market Structure. That was really when the SEC first threw its hat into the ring in any meaningful way in terms of getting involved in looking at all these issues. So hundreds of market participants responded to the SEC's Concept Release, which is what the intent was - to solicit public commentary on a whole host of structural questions, including high frequency trading and low latency, and direct data feeds and all kinds of issues. Our firm wrote one of the most detailed and, I guess, well-documented in terms of empirical research, responses to that and it has become a very widely cited source for information about high frequency trading. For example, if you look at the Wikipedia article on algorithmic trading, our Concept Release response is the most heavily cited source. That has nothing to do with us. We did not in any way, shape or form affect the content of that article. The point is it's been widely read and it's been widely read by the regulators.

They reached out to us to see how we did some of our analyses. They were very impressed with the tools that we had because we were able to do these analyses very, very quickly and efficiently, and I guess they had been looking for similar capabilities. So they put out a request for a proposal a couple of years after that and several firms responded. I don't know who the others were but we were awarded the contract last year.

Adam: Now they will be able to look at not just the trades, but the quotes and the orders. That makes a huge difference in terms of doing analysis of any given situation. Are there any findings you think will come out that might have seemed obvious to you or others in the market, but now will become more accepted?

Manoj: I think what's going to happen, what I hope will happen, is that the regulator will be able to take a strong stand that's informed by detailed research that it did itself, without having to commission out any research to biased sources. The regulator will be able to take a strong stand on all these issues that are currently controversial, and they'll be able to support their viewpoint empirically. And if that falls on the right side of what we believe, then great; and if it doesn't, then great. The point is that we're happy with any sort of data-driven decision making - that's good for the market.

Manoj Narang, Mani_Mahjouri

Manoj Narang, CEO, & Mani Mahjouri, chief investment strategist, Tradeworx

Adam: You mentioned at the start some regulations you'd like to see even though they might be to the detriment of your firm. Can you be specific?

Manoj: One of the policy measures that I've been advocating, you could even say evangelising, since 2010 has been a minor tweak to Rule 611 of Reg MNS which basically bans locked or crossed markets. I think the ban on locked markets is superfluous; I think the ban on crossed markets is fine. But the ban on locked markets serves no valid purpose. And apart from not serving a valid purpose, it has led to most of the problems in the equities market structure that people complain about. For example, the fact that we have 16 stock exchanges - in other words massive fragmentation of liquidity - that's a direct consequence of the ban on locked markets. The fact that we have all these exotic order types is a direct consequence of the ban on locked markets. The fact that you have such elevated cancellation rates in high frequency trading strategies is a direct consequence of the fact that there are so many trading venues, which itself is consequent to the ban on locked markets. My point is that virtually every adverse circumstance or adverse aspect of current equity market structure can be traced back fairly simply to that very superfluous regulation. My view is that the best way to go about decision making, not just policy making, but decision making in life, when it comes to complex decisions, is to pick the low hanging fruit first. This is a clear example of low hanging fruit. It would make the market better and eliminate all the unexpected consequences at the same time.

Another market structure reform we'd like to see is the elimination of rebate tiers. We think that rebates are a perfectly healthy aspect of the market. You can empirically show, using data, that they create deeper quotes and deeper markets. In other words, they foster liquidity. I'm not going to bother insisting on that because now the SEC has the tools to analyse that on their own I'll let them come to their own conclusion. But my point is that rebates are a positive attribute of the market, so long as they're fair. What's not fair right now is that the highest volume traders get higher rebates than low volume traders. That is extraordinarily unfair and we think that ought to be changed.

Adam: Those are private decisions, aren't they?

Manoj: No, they're not. The exchanges are only allowed to offer rebates because of Reg NMS. And that prescribes what the maximum level of rebate can be.

The fact is that merely posting orders passively into the order books of exchanges is a negative profit enterprise. You'll lose approximately 20-25 mils per share on average from adverse selection every time an order that gets filled. So market making on its own is an unprofitable enterprise. The only thing that makes it profitable to post orders in the order books of exchanges is the existence of the rebate. So, if some firms are getting a rebate that's 10 mils per share larger than others, that's the difference between being able to be profitable versus being unprofitable. Again, we think it's grossly unfair. And again, I'm not talking my position here. Our firm does get the highest tier of rebates. So when we're leveling the playing field that will not be to our benefit.