Haim Bodek, Decimus Capital
"The reality is that many of these practices, which were represented as secret sauce, were really just a broken vending machine. You hit it the right way and coins came out."
Haim Bodek, founder of Decimus Capital, talks to Automated Trader Editor Adam Cox about his campaign to shed light on the relationship between exchanges and a number of HFT firms. Along the way, he also talks about prospects for a return to more quant-oriented strategies and the implications of a volatility regime change.
Adam: You've been trying to draw attention to a lot of practices you think are unfair, even if some are, strictly speaking, legal. How much of an impact do you think you're having?
Haim: Prior to me alerting people to the crucial intersection between HFT profitability and HFT-oriented special order types at the exchanges, that interaction was not quoted on in any manner by the press. And now it's on the front page of many trade journals and it's been raised at a number of Senate Congressional hearings as well as the investigation of special order types which has been going on 18 months. In November of 2012, the SEC Office of Compliance Inspections and Examinations also launched a rigorous probe, which was expected to last a year, on the issue. On top of that you've got a number of structural changes at the exchanges. I don't want to go into details on what's changed - it's an ongoing investigation -- but I would say that of all exchanges, I've been particularly impressed with how proactive NASDAQ has been to disclose their functionality and also modify their functionality.
Adam: You're not talking about getting rid of HFT. You're talking about eliminating some advantages that HFT firms have had.
Haim: I think there's a theme here. I came from an algorithmic trading background, really a hybrid between microstructure technology-oriented trading and quantitative trading. What you see in HFT is the abandonment of the more quantitative traditions. And you see it focusing on exploiting loopholes. So I found the entire practice, once I got exposed to it, as being intellectually brain-dead. It's kind of an inside joke as to how brain-dead the strategies are. The reality is that many of these practices, which were represented as secret sauce, were really just a broken vending machine. You hit it the right way and coins came out. So I alerted regulators to the practices and I'm just watching to see how it unfolds.
Adam: There are plenty of studies that suggest HFT has brought down spreads. Do you feel there are some benefits to HFT?
Haim: I don't see anything inherently wrong with HFT trading concepts. The concepts make sense. I make a distinction. When I use the term HFT, I'm referring to the specific firms that got unnaturally large by playing specific strategies that exploited loopholes. And what you're talking about is the general tradition of HFT that has spanned the last 10 years. I would argue the term HFT didn't come into play in the greater financial community, or at least it wasn't a buzzword, until these specific firms which came out of nowhere became very profitable in a short amount of time, marketing themselves as being the next big thing.
Inherently, I have no problem with high frequency trading concepts. I have a big problem with undocumented loopholes at exchanges that are exploited to supercharge particular firms' trading profits.
Adam: Being a devil's advocate, if you're an HFT firm and the loophole is there, why wouldn't you exploit it? From your point of view is it more the exchanges that need to reform?
Haim: The issue here is that the features that I'm most concerned about were developed between exchanges and specific trading firms in a non-public manner. It's the features of the special order types, the specific conditions which are vital for these HFT strategies. One criticism that is now public is the queue jumping phenomena, which is a violation of price and time priority. That type of queue jumping behaviour is completely and absolutely unsubstantiated by any regulatory filing.
What HFTs know and what the rest of the world doesn't know, is that queue jumping is a feature that was requested: 'I'm not bringing orders to this particular exchange unless I can have this priority.' It really should be in the documentation.
Adam: Here you're referring to the practice of using hide-and-light orders, which you've talked about before.
Haim: Let me be very explicit here. NASDAQ rolled out a new version of hide and light which had specific prohibitions against queue jumping. So a hide-and-light order does not need that perk. That perk is hugely important, and it is undocumented on a number of exchanges.
I likely had some impact on that development. Maybe I'm too anti-social but I just don't want to get my information by having to go out for drinks with exchange marketing and sales representatives.
Adam: I was surprised to learn that you could have default settings for order sliding, and you wouldn't even necessarily be aware of it.
Haim: I think that's another thing. I've done a comprehensive investigation, a multi-year personal research project, into not only the features, but, when the material changes happened and who was informed. Were regulators informed? Did emails go out to participants? It was interesting for me. Because of that project, which I did mainly to prepare for this kind of public advocacy, I learned that a lot of things that happened to me earlier, when I was working at Trading Machines, went over my head because it wasn't properly disclosed. I'm saying, not only are your orders subjected to price-sliding treatment, but over the last five years exchanges have changed price-sliding treatment, impacting HFTs and customers without actually alerting people about that.
Adam: Would it require someone suing an exchange over what they felt they could prove were unnecessary costs or funds lost?
Haim: First they would need to assess the slippage and cost, which would mean that they would need to understand how the slippage and cost occurred. To me it's clear with the order-matching abuses that I'm concerned about. It is absolutely possible to go back through data and identify executions that were the result of inappropriate order handling. The slippage can be estimated from this data. But to actually do that analysis, you have to know about it. And right now, the only firms that know about it are the exchanges and a few HFT firms.
At some point I imagine there'll be some assessment that's made public, and then only after that could firms determine if there was any recourse.
Adam: Do you find you get some dirty looks from certain quarters at conferences when you go to speak? You've cast yourself as a whistle-blower and usually those people take a bit of flak along the way.
Haim: It's kind of funny. Of course you would think that the exchanges and HFTs would be the ones that were most angry about this, and not the victims, which are basically the buy-side and the major investment banks.
What happened was, the exchanges and the HFTs were actually quite friendly to me. I met with a number of the exchanges. I don't want to say that they were proactive, but they see this almost as if it was going to happen at some point.
The HFTs are a little bit different. There are a number of firms which I'm sure have a big issue with me. I thought it would be all of them, but actually there are HFTs out there where their attitude is more like, 'All right, you got me!' It's more like I actually get some respect in that I was a non-insider and I clued into their game.
The guys I get along with the most are actually some of the middle-of-the-road HFT guys. And those discussions, a lot of it is oriented towards: 'Okay, the old game is over, where do we go from here?'
Adam: That's an interesting issue, because a lot of the market-making activity which leads to the scalping which a lot of firms engage in, is not inherently profitable without some of these conditions. So, to the extent that HFT activity provides a benefit, how do you strike a balance? How do you have it so that you have the 'good HFT', that allows for market-making and narrow spreads and lots of liquidity?
Haim: Now you go back to your traditional algorithmic market-making concept, where automated market-makers have obligations and they have advantages - and those advantages are completely understood - and they aggregate customer volume.
I don't actually have a problem with markets constructed with asymmetry. I have a problem when markets are constructed in a way that is discriminatory, in a way that is not public and potentially collusive.
Adam: So it's like the difference between knowing the odds and the risk you might take versus being led to believe that the odds are different than they actually are.
Haim: It totally makes sense to me that it went to this place, when the function of the market-maker eroded. The investment in a market-making operation is significant. But what you're also going to get is a race to the bottom. You're going to get compression. Decimalisation assisted that process.
You know, it's extremely difficult to trade 5% of a name, for paper-thin margins, in an electronic marketplace where all the toxic flows are being routed to the places where you trade. The main lit exchanges in the US are receptacles for all the trades after internalisation dark pools cherry-pick all the neutral flow. So you have a systemic issue.
One HFT I talked to said to me, 'You know, we can't make money without these features.' And he's kind of right. The US marketplace is constructed in a manner right now where, without that unnatural edge, it's very unclear how market-makers are supposed to make money in the electronic screens. Everybody knows how to make money in an internalisation, but being out there in the screen, all day long and making markets in these stocks is incredibly difficult.
Adam: Is there a danger, do you think, as these issues get addressed, that spreads will widen again? There's always a balance between competitive forces pushing spreads down and the need to be profitable, so if HFT firms became unprofitable, what happens?
Haim: I'd argue a lot of HFT volume is kind of artificial. It's basically interpositioning of HFTs between natural buyers and sellers. I have a friend, a quant, who was asking me: 'Why do we need market-makers in these highly liquid names?' We've seen in the Flash Crash that this type of liquidity doesn't stay there. There's always an argument for why we need to have a market-maker. But these market-making strategies, if you look at them in practice, they're just serving to tax access to the market.
I like stat-arb strategies because I would argue that those are more aligned with the idea of providing depth in risk capacity in terms of liquidity. High frequency, intraday, back-and-forth, end-of-day-inventory-neutral strategies, I don't really see them as having to be protected.
If they go away, I'd argue you'd have a rebirth of quantitative finance. There would be many strategies that I was exposed to in the early 2000s/late 1990s that probably should be still operating but were actually side-lined by the HFT technology-driven area.
So I'd like to see intraday algorithmic quantitative trading return. The other thing is, institutions would be significantly more active in the screen if they weren't constantly exposed to the adverse selection of HFT, which is related to the 10-year-old constricted tradition whereby exchanges have catered to that client.
What's supposed to happen is, market-makers are only supposed to step in when liquidity is required. They're not supposed to get in between two customers who are going to collide in the next 10 seconds. Those customers I think have run away from exchanges to dark pools, and I think there's an opening for them to return.
Adam: I'm interested in the idea that some quant strategies could come back into vogue.
Haim: Well, there's some compression in the stat-arb space after the liquidation crisis of August 2007. A number of them were shut down. It was actually in the same period that Reg NMS came in. I've spoken with extremely sophisticated people from the stat-arb space, who remind me of how difficult those market structure changes were for their strategies. They were trading for paper-thin edges and Reg NMS comes in and all the special order types, and that side-lines the stat-arb space from the market-making function it was performing in the first half of the 2000s.
I don't actually think the knowledge has been lost. I just think that the economics and the slippage and the fee structure have basically made it more difficult for those types of firms to act like liquidity providers.
Adam: So a firm might come up with interesting models that factor in a wide variety of markets and stat-arb opportunities, only to find that there's no money to be had.
Haim: Think about how many times you've heard a strategy's in-market performance being dramatically different from its back-test. I doubt there are more than like 10 quants out there who can develop and back-test a sophisticated stat-arb model and at the same time understand all the execution nuances linked to these exchanges. It doesn't matter how good your model is. If you put orders out there that are heavily biased, where you're just trading the adverse-selected trades, you're just never going to have enough alpha to overcome that adverse selection that is part of the playing field right now.
From what I hear, that's what they (HFTs) are trying to do, they're trying to get more into the stat-arb space. I'm surprised there's no stat-arb firm that's going into the HFT space in scale.
Adam: It's an interesting point. Within the stat-arb space, what's the sort of time horizon we're talking about? As you have a multi-market position, how long does it usually stay open?
Haim: I've been exposed to the market-making strategies in the first half of the 2000s that were successful that had a stat-arb approach. From what I understand, there's kind of a sweet spot of around 15 minutes.
It was surprising to me that, it doesn't seem like those strategies have been married with this ultra-low latency tradition.
Adam: There's been a lot of evidence that suggests the HFT industry is less profitable. As you try to go into longer time horizons, what are the challenges that you face?
Haim: I actually see a cultural issue. If you look at the top HFT firms, you're going to see traders who are ex-floor traders and they're going to know the microstructure pretty well. If you look at the top quantitative trading firms, many of them have had quant legends, who may not have traded on the floor in the Chicago pits. What I've noticed is that there's a huge cultural divide between quantitative trading firms and HFTs.
When HFTs first started appearing I was a bit confused because they weren't hiring quants. They were approaching developers and some traders, but there weren't any quants on the headcounts. A lot of people think that these firms have back-tested the data when it's totally not the case. A lot of the strategies are developed in the production system, trading one share at a time. There really isn't a strong financial engineering backbone to these types of firms. And I would also argue that the top quant shops actually are pretty poor in terms of the technology competency.
Many are structured as hedge funds. If you're not a firm that has to fulfil market-making obligations, you just don't need a as much of a tech team. A market-maker has to have a very strong tech team.
Adam: We've been talking about these big issues but we haven't talked about Decimus and what you're up to. I think you've mentioned you're looking at options?
Haim: I've taken on this role, kind of a regulatory-reform role, and before it was public the industry knew I was going in this direction, so it really didn't make sense for me over the last few years to be working in an investment bank. This is a road not travelled, for someone of my background.
My firm does consulting on special order types and how to protect yourself from abuses. And I give a lot of talks about that. But the project I'm most excited about now is a project to get back into the options space without having a $10 million infrastructure build. I'm fascinated with the idea that there can be high-volume options-oriented strategies that can trade with high frequency without necessarily having to do low-latency.
Adam: One final question for you, as a market practitioner. One of the biggest issues we're having is the gradual ending of the massive amount of monetary easing we've had. We know we've had an abnormal period for monetary policy, so we're going to be undergoing a market regime change. How do you play that, or how do you advise someone to play that?
Haim: I'm as high-volume, intraday as it gets. The period of low volatility we're having right now is just killing the opportunities for intraday quant trading and market-making. So I'd say over the longer run, I can't say what people should play, but I can say that whether you like it or not, you're either long or short volatility in your business.
That's the big question: What is the market regime and volatility going to look like over the next five years?
Of course nobody could have expected the previous five years. The successes and failures of algorithmic trading businesses have been hugely impacted by the volatility over the last five years. And your decision of where to invest and what hand to play over the next five years really comes down to either a volatility bet or a volatility hedge.
I'm not thinking of a macro fund investor. I'm saying, if you work at an algorithmic trading firm, the way that firm copes with both regimes, higher volatility and lower volatility, will impact your career, whether you're laid off or whether you'll do well. You know, three years ago, the idea that HFT would collapse like this was unfathomable. The key thing is that we have volatility at historical lows.
You're exposed, and how you or your firm hedges that exposure or manages it, will actually have the most impact on your pay-out. I mean, how many people overinvested in HFT and are sitting there in this completely perplexed state?
1995 - Earns Bachelor's degree in mathematics and cognitive science, University of Rochester 1997 - Joins Hull Trading in financial engineering group, focused on options algorithmic trading strategy
1999 - Hull Trading is sold to Goldman Sachs, Haim remains at firm as a financial engineer and electronic trading strategist
2002 - Co-founds and briefly runs software and infrastructure firm Sixfold Technologies before joining UBS
2003 -Sets up US electronic volatility trading business at UBS, later becomes global co-head of division
2007 - Founds Trading Machines to build an HFT options trading strategy for the shift to decimalisation in US options markets
2009 - Starts investigating HFT practices in exchange order matching engines
2012 - SEC investigation of HFT special order types provided by exchanges becomes public issue