Damp squib (expression, British
1) A miniature explosive device that has become wet, preventing it from working;
2) A disappointment;
3) A description of what trading volumes in the equity markets were like in the past couple of years.
Is it any surprise that firms have ventured off in search of new pastures, trying out different asset classes, given the persistent lull in trading activity in the world's equity markets?
Since late 2008, interest rates in the United States and many major economies have been at rock-bottom levels and have shown scant sign of rising. But instead of sending investors flooding into equity markets and spurring strong volumes, the near-zero rate environment has led to subdued and somewhat orderly trading. Some of this torpor can be attributed to a post-crisis hangover, with general investor confidence still fragile after such an extreme, once-in-a-generation event. But market watchers say it is also clearly a function of the low volatility that stems from the ultra-stable interest rate outlook that has been signalled by central bankers.
Faced with such an environment, many trading firms have chosen to widen their horizons and trade new markets. A study by Barclays Capital that questioned 1,300 institutional investors found 85 percent of respondents were responsible for multiple asset classes. That compared with about 50 percent when the survey started in 2005.
In Automated Trader's own survey, which featured more than 650 buy side and sell side participants, more trading was expected in every single asset class listed. In the main cash markets, growth was expected to be most muted, with 53 percent now saying they trade cash equities and just 55 percent expecting to in two to three years. But in several derivatives markets - notably equity options, fixed income derivatives and foreign exchange futures - the share of firms expecting to trade was some 9-11 percentage points higher. All told, the survey highlighted the appetite among trading firms for taking on new challenges.
Still, conversations with market participants, investors and technologists suggest that the business of wading into new asset classes is no simple matter. From mundane, nuts-and-bolts factors such as connectivity and due diligence, to more complicated tasks such as reconstructing models, firms find that trading or selling products for a new asset class can bring up a host of issues to consider. To that end, we set out to hear from executives, quants, academics and others just what those issues are and what you need to think about before diving head-first into a new market.
This year's model
At the heart of any transition into a new asset class is the task of coming up with new models.
"Every market has its own idiosyncrasies, different exchanges, different trading strategies, different supply and demand factors, that make it somewhat difficult," said Scott Morris, head of quantitative research at Ronin Capital in Chicago. "So you have to tailor your models for every market."
(To see the video interview with Scott Morris, click here.)
"Every market has its own idiosyncrasies, different exchanges, different trading strategies, different supply and demand factors, that make it somewhat difficult."
Morris said coming up with new strategies and models is harder than people often realise because they inevitably come with a set of biases based on their previous experience.
"It takes longer than people think. And one of the reasons it takes so long is that you really never know until you start trading because some of the assumptions you had in the other asset classes, you perhaps didn't realise they were constraints. But once you start applying them to a different asset class, those assumptions catch up with you," he said.
One example of the kind of tripwire that a trader may face is the liquidity profile of a given market. If you're used to one kind of liquidity, say, in blue chip equities, you may find life very different in commodities or fixed income.
"That's a perfect example of a hidden assumption. You develop a model that works in a high voluminous world or asset class and then you go to one less so, you have that assumption built into the old model and you have to then tailor your model," he said.
Even moving from a cash market to derivatives within the same nominal asset class can bring risks.
"There's a number of examples of equity firms moving into the equity options space a couple of years ago, where they just saw options as another source of delta for them to put their trades on and it just didn't work out for them. Of course the timing perhaps wasn't great for that move as well, but that's a good example," Morris said.
Linked in markets
The model-making challenge is not only about accounting for how one asset class behaves differently from another on its own, but also about how it interacts with the wider market.
"You always have to be aware of the fact that the markets are linked," said Rick Cooper, an assistant professor of finance at the Illinois Institute of Technology.
(To see the video interview with Rick Cooper, click here.)
"You always have to be aware of the fact that the markets are linked."
"The classic example is that many low frequency firms had much information to add to the economy, but they traded once a month, because every time they came into trade, they widened the spreads and they couldn't afford the slippage. Now, with high frequency, some of them can trade every day, because the high frequency traders add liquidity. So that affects the information flow, it affects the value. It also affects how long the signal is valuable."
By the same token, he said, derivatives markets will affect how long a stock signal can remain valuable. "So you can build your model sort of in isolation - 'I'm building a stock model', 'I'm building a bond model', 'I'm building a fixed-income model' - but how well that model works will be contingent on the dynamics of the other markets. And you have to take that into account when you're doing your analysis of your trade. As you start to see things slip, it isn't always just your specific market that's making it slip," Cooper said.
Cooper is not your typical chalk-wielding professor. For a start, he spent years in the markets as a portfolio manager and director of analytics before deciding to go back into academia.
He also said that when he's teaching students, he doesn't worry so much about asset classes.
"When I teach a class I try to give them a more basic approach, which is to say: What is it that makes money? Doesn't matter which asset class it is, you're either going to make money because you have a better statistical forecast, you have better information, or you have technology that's an advantage in the trading," he said.
"In the end, we're trying to produce a product that our employers will want. And being in Chicago, the product we think employers will want are people who have good applied skills, that will work in … high frequency or derivative markets," he said. "But we don't know if they're going to get employed by the CME or a derivatives trader or by GETCO, so they have to be kind of a Jack-of-All-Trades. They have to know how to think about the problem."
Raphael Markellos, who is chair in finance for the Norwich Business School at the University of East Anglia, also said asset class-specifics did not really enter into the discussion with his students. The idea of a university education was to learn about principles, approaches and concepts which could be applied to a whole variety of problems, he said.
"The principles are the same or the tools can be the same, but obviously different data sets or different asset classes will have unique characteristics, and you need experience, you need to define the data sets and the problems that they have, the noise that's involved," he said.
In commodities, for example, electricity or energy markets, seasonality can be such a key part of the data. Another factor could be the frequency and scale of extreme observations, as well as the question of whether to even include outliers in a dataset. "So in electricity data, you'll see a lot of these extreme observations. They're part of the normal behaviour of the data."
Markellos's own research spans an extremely wide variety of datasets, from wine to weather to football scores. He describes himself as "a methodology person" who views different datasets as challenges. "If it's a mountain, we have to climb it," he says.
As such, he has one piece of advice for anyone building a model in any asset class. "One very important lesson - which I learned also from my teachers - is that before you start dealing with models, you have to really, really prepare your data properly, to clean them and make sure you've got a good dataset to work with."
He said many people ignore this step as they already have an end result in mind and are less concerned with how they will get there.
Nuts & bolts
So you've spent a lot of time preparing your data and you've taken care to challenge your built-in assumptions when it comes to your trading model. What next?
Mike Madigan, the chief technology officer at WH Trading in Chicago, knows a thing or two about making it possible for his firm's traders to go after new markets. WH Trading has about 75 people trading a long list of products from interest rates to agriculture. With three data centres in Chicago, racks in Singapore, Tokyo and London, the firm is active on the CME, NYMEX, COMEX, Eurex, LIFFE and other venues.
"Some of the interest rate products historically have been some of the key products for us. And with interest rates being so low and with volatility so low - the Fed saying it's going to be this way until 2015 - yes, we have had to start looking for other venues to trade in," Madigan said.
First off, there are technical issues such as making sure software is set up for new contract specifications. That can be a question of ticks versus decimals, but generally speaking he says those changes don't require massive software alterations.
More challenging is the broader due diligence required to go into a new market.
"First of all, anytime we want to go to a new exchange, we have to do some analysis on what do we think the opportunity is. The trader has to look at the marketplace and say, 'How much volume is there, how much volatility is there, is this even worth going to? Can I bring an edge?' And then, they'll sit down with me and the owners and say, 'Look, here's how much it's going to cost - it's not free - to go to any of these exchanges.'
"It's a fairly expensive proposition, just from a one-time cost of going there," Madigan added, noting that on top of that is the monthly cost of connectivity, market data feeds, routing fees and other costs.
"So there are barriers to entry, both capital wise and frankly just pain-in-the-neck wise, apart from the costs that it's going to take you to do it in the first place," he said.
All told, the time from the moment a trader decides it's possible to make money in a new market to when the firm is set up and ready to go can take six months, Madigan said.
But there could be deeper issues. Adjusting software for new specifications or setting up exchange connections and feeds may be relatively straight-forward. But trading different, or multiple, asset classes can affect your overall technology infrastructure.
Simon Garland, chief strategist at Kx Systems, said the kind of database a firm uses can become critical.
"Part of it is just simply the speed," he said. "If you've got the speed, that means you can afford to be taking transactions coming from one place or another which have, in database terms, different schemas, different representations."
The differences may be slight but in a millisecond-critical world that still needs to be addressed. "If you can have the speed to be normalising on the fly, that suddenly means you have proper risk management over the asset classes," he said.
Kx's database technology is asset-class neutral - in fact it began life as a programming language which clients began writing databases for.
Finally, however, there is the moment of truth. Morris of Ronin said that only when you start trading a new market does it become clear if you did enough homework. "You can back test as much as you want but really the essence is getting in there and finding out the trades. So it's a lot harder than most people think."