HFT Case Study – Allston Trading
Thu, 07 Jul 2011 10:46:00 GMT
An Interview with Peter Nabicht
In this interview for the High Frequency Trading Review, Mike O’Hara talks to Peter Nabicht, Chief Technology Officer at Allston Trading, a proprietary high frequency trading firm headquartered in Chicago, IL.
Peter joined Allston in 2004, began a state-of-the-art real-time operations desk to support trading activities across multiple asset classes, headed up technology on the Short Term Interest Rates, and became CTO of the firm in 2008.
HFT Review: Peter, could you give us some background on Allston Trading?
Peter Nabicht: Allston is a principal trading firm that is engaged in some high frequency trading, but like a lot of principal trading firms that do HFT, it is not our only focus. We started out primarily because trading on the exchange floor was too slow, there were times where you would be trading in the pit, trying to let somebody upstairs know what you did so they could hedge your position, but by that time it was too late. So when the CME first started to go electronic, our founders started up the company and now, eight short years later, we are on fifty-plus endpoints around the world and we’re in almost every asset class.
Just like a lot of companies similar to us, we like to be centrally cleared. We like minimal counterparty risk and we like electronic exchanges with central limit order books. So we really don’t do much OTC business. We trade options, equities and primarily futures.
We started in and work primarily with futures. It’s funny, I’ve been reading all these articles in a variety of publications about high frequency trading, saying the next big thing in HFT is futures and I want to call up the authors and tell them “Futures has had HFT for years!”
HFTR: Would you classify Allston as a market maker?
PN: We do make markets but there are a lot of misconceptions out there as to what a market maker actually is. You have situations where you’re making markets on one product on one exchange, you get filled so you have to go to another market to get your hedge off. Just because I’m acting as a taker in my hedge market doesn’t mean I’m not a market maker on another. As an industry we have to stop looking at products and markets and asset classes as unconnected entities and start looking at the whole of the global market as the interconnected system that it is.
So I prefer not to classify Allston as solely a market maker – even though we do make markets – because we occasionally take them, whether for hedging or for some other reason.
Primarily we see ourselves as liquidity providers. We might not always be on both sides of the book but more often than not we’re resting orders. If we have a spread between products for example, we like to rest one side of the spread in each leg, but that might look like only one side on each market, which will immediately make us not count as market makers! It’s hard to classify so I’d say “liquidity provider” is a better way to put it.
HFTR: You said that Allston does a good deal of high frequency trading. Can you give us a quick definition of what you would class as high frequency trading?
PN: I like to classify high frequency trading as strategy-independent. I think the best definition of high frequency trading is putting on and taking off positions so rapidly that it’s really only made possible by using current technology.
A lot of people, whether it be the press, regulators or legislators like to talk about “the high frequency trading strategy”.
But HFT is not a strategy; it’s a means of execution. It’s not a trading strategy, it’s a business strategy. It’s how you want to go about doing business, how you want to execute your trades and how you want to go about making money, but it is not in and of itself a trading strategy. That would be like saying, “that pit trading strategy”, or, “that long term investing strategy that all those mutual funds run”. None of those are accurate statements.
HFT is a way of running any number of strategies. In fact, any company engaged in HFT worth its salt runs scores of strategies. Some of which may be considered high frequency and some not.
At Allston, we’re always looking for new strategies. I was recently at a US university, talking to a group of kids coming out of graduate school who were looking for jobs in the industry. I suggested to some of them that they should look at becoming quantitative analysts for high frequency trading companies. And it surprised me how many said things like “well I haven’t been doing that because they have their strategy and they already do everything so what role is there for me?”
We need to educate people that there isn’t just one strategy here. This misconception will go away the more those engaged in HFT talk about it in straight forward and meaningful ways.
HFTR: But would it be fair to say that some strategies are only possible through HFT, that there are some strategies that you would not be able to deploy unless you had a HFT infrastructure?
PN: Yes, I think the best way to describe it is that high frequency trading lends itself to a lot of strategies, that it improves them. It helps you find an edge within a strategy that may or may not otherwise be possible, but it’s no more influential on a strategy than the market structure for example.
Actually, low latency is really what makes some trades possible. And a lot of companies wouldn’t go down the low latency route unless they were also doing some high frequency trading. Developing a low latency platform is a huge investment of time, people, and money. In order to do so you need to have some consistent profitability from strategies you run, whether HFT or not. It just so happens that HFT requires low latency, so you see the two go hand-in-hand a lot.
But speed isn’t always the most important factor. I remember running a trade for a while on the energy desk, a very basic spreading trade, between the physical and the cash delivered crude. We automated a trade that one of our guys was running by hand (point-and-click). Then we ran them side by side. Guess what? He was consistently making money when I was losing money. When we looked into it, we discovered he was actually making money because he was slower.
HFTR: How could that happen?
PN: Basically the automated trade was to immediately process every tick of those choppy markets and it was dumping position at our slippage tolerance. But when trading manually he wasn’t able to see every tick update and react to every little move so he was missing the market going against him and getting filled on his hedge when the market was coming back. He wasn’t dumping as quickly, and that was working to his advantage. Some might say he was more risky because he wasn’t reacting fast enough and some might say he was more profitable because of it.
Getting back to the point, low latency makes some strategies possible, but it isn’t a necessary expense for other strategies. I think it’s just how people want to focus on trading. I refer to pit trading a lot to help point out that high frequency trading, or low latency trading, is just the next step, or at least part of the next step. High frequency trading doesn’t make trades any more possible than being on the top step of the pit and having a louder voice makes certain strategies possible.
HFTR: What is your typical holding period for a trade? And do you ever hold positions overnight or are you always flat at the end of the day?
If you are not a registered user, click here to register for free