Youngju Nielsen is the co-founder of Quantavium Capital, a fixed-income fund based in Jersey City, N.J., that will begin operations in the third quarter of 2012. Nielsen has 12 years of investment experience in the fixed income space, running global interest rate portfolios at the proprietary trading groups of Bear Stearns, J.P. Morgan Chase and Citigroup. She built the proprietary quantitative strategy for the fixed income prop trading desk at Bear Stearns, as well as the systematic rates trading business at Citi. Previously, she was a research officer at Barclays Global Investors.
What is Quantavium Capital's investing model?
Nielsen: We perform statistical arbitrage on global fixed income -- major countries' interest rates -- using quantitative discipline and a systematic process. We trade the most liquid instruments: futures, cash notes and bonds. You might have seen similar approaches in other asset classes, such as equities, but this is a relatively young and fast-growing area in fixed income that benefits from recent years' developments in electronic trading.
How much do you have in assets under management?
Nielsen: We expect to have just under $50 million at launch.
How many traders do you have?
Nielsen: Three traders. We outsource most of the functions other than research and trading. Our hiring will be relative to our capital inflow.
How did you select your team?
Nielsen: The team members have previously worked with me, especially our senior algorithmic trader, who was in charge of development for my systematic trading business. He is also a partner of the firm. He has 12 years of technical and trading experience in fixed income algorithmic trading.
Are you a high-frequency trading shop? Do you use algos?
Nielsen: Our holding period is anywhere between 1 second and three days. We are not an HFT shop, but we still need to react to market movements as quickly as possible.
We trade algorithmically. Our daily trading activity is entirely process-driven, and a set of highly co-related processes controls all trading activities. Real-time market data and historical data are fed into various models to calculate alpha models as the trading schedule starts. When the signal is triggered, the algorithm generates the desired position, orders and baskets. Before routing baskets for execution, the system will cancel/combine orders and decide order types, based on the current market situation. We also have systems that provide traders with real-time execution and transaction cost feedback.
Where do you get your algorithms?
Nielsen: We developed most of our algorithms in-house. Our algorithms can be divided into three categories: alpha-generation models, risk models and transaction cost models, with some optimization to combine these three inputs.
The alpha model gives predictive returns. Each alpha model can respond to a combination of multiple signals.The risk model defines common factor risks, specifically yield curve risk and idiosyncratic risk, and takes account of the relationship among different securities. Finally, transaction cost is factored in.
What types of algorithms do you use?
Nielsen: Because we're trading fixed income, we often use algorithms related to fitting the yield curves of different countries. Then we construct various types of statistical models using market data, historical data and other related data. We process this data and tick data using very specific filtering methods.
Our risk model is constructed around fundamentally driven quantitative methods, rather than pure statistical methods. We use an optimization algorithm to decide the position size.
Which electronic trading platforms do you use?
Nielsen: At Bear Stearns, we started trading more through BrokerTec or eSpeed, where you can see a lot more numbers and market depth from different players, as opposed to looking at just one broker's inventory or requests for quote through TradeWeb or Bloomberg. However, our trading focus shifted to futures when we started Quantavium. We found it is convenient for a start-up to trade exchange-traded securities, and it seemed to be more attractive to our prospective clients.