Buy-side firms are notorious for buying everything, from technology to their algorithms. Algorithmic Insight Capital Management of Clearwater, Fla., run by Julie Connard and her husband, Mark, is the rare exception, however. This high-frequency quant fund builds its own algo formulas, trading platforms and research tools. And the Connards say they love volatility.
Describe Algorithmic Insight Capital Management.
Julie Connard: We're a hedge fund. In the past we were more of a prop-trading firm. Up until very recently we did not manage money for anyone else, just our own firm's money, and we decided to start a hedge fund. We have our own asset management firm called Algorithmic Insight Capital Management and we have a hedge fund called Next Wave Capital Partners.
Is your firm a long-hold or short-hold fund?
Connard: Usually quite short. It depends on what strategy we are using. Some of our strategies can be long, but we are fairly high-frequency and we are oriented toward short trading.
You and your husband, Mark, run both firms. What assets do you trade?
Connard: Right now we trade all U.S. equities and options under equities. We will be branching out to other asset classes eventually because our algorithms can be applied to other asset classes. We will most likely be branching out next into futures. However, before we branch out to other asset classes, we will be doing a lot more work on automating our options trading systems.
Are you updating your existing systems, or is this a complete technology overhaul?
Connard: Right now our automated trading system is designed to trade equities. That complexity of trading options is fine when you are doing it manually -- that's what expert options traders are for. But when you do automated trading of options, in terms of the technology involved, it's very complicated.
We actually have built all of the infrastructure we need to do automated options trading already, and that infrastructure is really quite extensive. We spent a good year building this infrastructure. When you deal with options chains, you have a lot of complexity. We are not actually doing automated options trading yet. We devoted a lot of time to preparing our systems, but we have not yet done the programming to utilize the very complex strategies that we use for options.
What percentage of your volume is electronic?
Connard: Essentially all of it. We will do test trading manually, but everything we do is essentially all automated and algorithmic. But when you are automating trading strategies you have to have a very good feeling for what you are doing. So we have a couple of ways we prepare for automating a strategy, including test trading, and sometimes we do that manually. We also do simulations -- we have an advanced simulator that we invested in.
With options strategies, we often do manual test trades. They are real trades, but they're not intended as a way to make money. It's a way of working out the exact algo we want to automate; it's preparatory trading. But in equity trading, we do very little manual trading.
Where do you get your algorithms?
Connard: Everything we do we develop internally. We don't use any third-party algorithms. We simply have absolutely no reason to use anyone else's algorithms.
We have been doing this since 2001, and back then you didn't have other firms offering algorithms commercially -- at least not to smaller shops. Starting as we did in '01, at that time we had to develop everything ourselves. We simply never found any need for any third-party algorithm.
Do you develop algos yourselves, or do you have a basement full of Russian mathematicians?
Connard: [Laughs] Well, we don't have a basement full of Russians, but we do have software engineers. My husband and I work out all the trading strategies and we work out the algorithms at the top level. My husband, by the way, used to be a senior software engineer at IBM. Then we take the strategy and the algo and we give it to one of our senior software engineers -- Chuck Prenner, who has a degree from Harvard in applied mathematics. We give it to him and he then writes to specification and gives it to a senior software engineer who does the coding.
We also have a new research arm. We hired another Ph.D. who is working on the research analytical end. He is not doing production trading programming, more research writing.
How often do you update your algos?
Connard: All the time, constantly. We're continuously updating our code. As for how often we have a new release, we don't have a set schedule for new releases. We usually do several new releases a week.
How many traders work for you?
Connard: Because we do automated trading, we don't need traders. The computers do the trading. My husband supervises all the trading; so I guess Mark is the chief trader, because he supervises the machines.
What trading platform do you use?
Connard: It's all systems we built ourselves. We use our own proprietary OMS.
What has been your biggest IT project over the past year?
Connard: It's difficult to say, as we are always programming new things and overhauling new systems. We have developed a new long/short dollar balancing system that allows us to maintain a very exacting dollar balance for long and short submissions. That is one large thing we have done. We can be exactly dollar balanced -- literally to the dollar.
We are excited about the work that our research Ph.D. engineer is doing. We are developing new predictive algorithms that are delivering good results. We are using new technology to do that. In the past we have not used predictive algorithms; we have been more a market-making and arbitrage shop. This doesn't require prediction. Now we have two Ph.D. engineers working on this, and we are very excited about it. We also are working on a new ETF arbitrage system. We are very busy.
And by the way, we are a distributed company. We have Chuck Prenner in New Jersey and our other Ph.D., Ryan Taylor, is based in Los Angeles. Not everyone is in Florida with us.
Do you use dark pools?
Connard: We don't. We may in the future, but right now, no. We just haven't really needed them. But we may in the future.
Do you use transaction cost analysis?
Connard: We measure transaction costs directly in our own software. Once again, we may in the future use third-party solutions, but because we build everything ourselves, we can get that data at a very, very high level.
How are you handling the current volatility?
Connard: We actually like volatility. [Laughs] So we don't mind the current volatility at all.
Our two most profitable trading days ever were the two days at the height of the financial crisis. One was in 2007 and the other in 2008. And we just made gobs of money on those days. So volatility is fine with us.
What about the May 2010 Flash Crash?
Connard: We got out of the way. Basically, our algorithms didn't like what they saw, and they got out of the way. Phil Albinus is the former editor-in-chief of Advanced Trading. He has nearly two decades of journalism experience and has been covering financial technology and regulation for nine years. Before joining Advanced Trading, he served as editor of Waters, a monthly trade journal ... View Full Bio