We have other risk models that are more predictive in nature where we look at tracking error and total risk at the individual stock level to make sure that our risk expectations are in line with what our clients expect.

Q:  How many holdings do you have at any time and what is your turnover?

A: Our holdings are generally about 70 to 100 long and 20 to 50 short, with net market exposure equal to 100 percent. Turnover ranges between 150% -250% annually.

Q:  How is your research process organized?

A: We have a research staff that we call the advanced research center. In our Boston location there are around 30 research analysts. These analysts conduct quantitative research. We have three research analysts within that advance research center who work exclusively with our U.S. active group and I’m a member of that team. This portfolio is managed through this team.

It is very collaborative in nature. The portfolio managers on the active team may have research ideas on how to improve the models based on observation in the marketplace and oftentimes we are able to discuss these ideas through this research team. Often, the research folks will have research ideas based on academic research or models from other parts of the company or other teams, and they are shared by the team as well.

We meet every morning to discuss news and corporate events of the day. We also have scheduled staff meetings every two weeks to go over research projects that everybody is currently working on. Then we have a higher level meeting every couple of months which defines our research agenda where everybody brings their best ideas for future research, and we talk about priorities and which ideas we are going to work on.

Q:  What are the things that make you unique with that research capability?

A: Unlike a lot of quantitative managers, we are able to do things that are much more difficult, intensive and mathematically rigorous because we have quantitative focus and designated research people. That gives the distinct advantage, particularly in the U.S. market where there are large number of companies and heavy flow of news and information.

Q:  How do you implement your research ideas?

A: We use our researchers’ and portfolio managers’ ideas to screen out ideas. We test them thoroughly across a number of time frames and metrics. We examine them in a real world setting and the portfolio managers put their personal check on the idea - should this work, does it fit with our philosophy, are these things driven by investor behaviors and are they likely to persist going forward.

Once the research idea meets those criteria, we will go through a rigorous process in finding a way to incorporate that idea, which takes a form of a quantitative model. We find a way to incorporate that new model into our existing model - we back test it to understand the stock behavior during different economic cycles and investor behaviors, we simulate it in a portfolio setting and we make sure the risk adjusted returns are in our favor.

There is one further step. Our in-house committees have to approve any significant changes to our models. We have an investment committee and we also have a technical committee which is staffed by senior people. We don’t change these models on a daily or a monthly basis. Any change that we undertake is driven by intensive research over months and these changes are reviewed by the committee. This keeps models and the investment process stable, but at the same time, we are adding value through incremental changes that have undergone rigorous testing and approval.

Q:  How does an idea turn into a holding?

A: We are always looking at how our model is performing and how we can improve it. For example, we noticed that at industry level we were not doing a good job in the semiconductors industry so we decided to see if there is a better way to do this. We undertook a significant research project for two reasons. First, there are a lot of semiconductor stocks out there. Second, there is a lot of volatility in these stocks. If we can get them right, we’ll add a significant amount of value for our clients.

So we undertook the project and we used an advanced mathematical technique involving quadratic programming and some other evolutionary techniques. Out of that came a number of factors which drive these stocks that we would not normally expect in semiconductor industry. These techniques allow us to not only identify new factors, but identify ways that factors interact with each other. Thus we established a semiconductor model that ranks only semiconductor stocks. We ran it by technical and investment committees. It was approved, we implemented it in our financial models and it now is part of our model. So every day, this model picks all the semiconductor stocks in our universe and ranks them from top to bottom in terms of expected return based on the factors we identified, and we use this model in picking stocks. After this model was implemented, we did some trading and bought stocks that looked attractive on the basis of this model, and we sold stocks we owned that were not attractive in this model.

Q:  How do you harvest gains and losses?

A: Our trading is driven exclusively by the risk reward criteria in the portfolio. Every day we are looking at the portfolio level and if we were to trade, what would be the expected return of the portfolio. We do not trade explicitly due to a gain or a loss, more typically we will trade because the outlook for the stock based on our models has changed significantly.

In the case of the semiconductors example, some of the trades that we may have done are based on the risk-reward relationship we see in the future, and sometimes they are based on a gain or a loss in the stock.

Q:  What is the goal of the fund and do you measure returns against any benchmark?