Q: How did the fund evolve?
A : PNC Multi-Factor Small Cap Core Fund was launched on September 30, 2005 and currently the fund has about $34 million in assets. Since the launch of the Fund, Paul Kleinaitis, my co-portfolio manager, and I have worked together on strategies focused on small cap investing. We refer to the team as the Structured Equity team and most of our work is based on quantitative discipline and systemic investing principles.
Q: What core principles guide your investment philosophy?
A : Our investment philosophy rests on three investment tenets -- reasonable valuations, improving fundamentals and investor behavior. We look to invest in stocks that are trading at reasonable price to its history, to its peer group and to its sector groups. In addition, we look for companies with improving earnings and business fundamentals along with company performance. And lastly, we look for favorable investor behavior measured on technical factors like price or market-driven factors such as volatility, liquidity, or momentum. We believe there is a direct relationship between investor behavior and the price of the stock. We don’t invest in stocks that are neglected or in stocks that are trading at sky high multiples. Another important facet of our strategy is using state-of-the-art technologies to capture market inefficiencies that our proprietary models detect.
Q: How do you translate your investment philosophy into the investment strategy?
A : We invest primarily in stocks of small cap companies that possess both value and growth characteristics. We research market data based on various factors that we track including fundamentals, valuation and investor interest factors. In the fundamentals bucket, we focus on earnings diffusion, earnings dispersion in the forecast of the analyst, sales growth, earnings surprise, income surprise, net margin change, sales outstanding, and approvals. Valuation metrics look to earnings-to-price, forward earnings-to-price, sales-to-price, cash flow-to-price, enterprise value, EBITDA and book-to-price of a company relative to its return on equity. Then we take individual factors, test them in different scenarios and determine which appear to be optimal for the small cap growth universe and build an eight-to-nine factor model in a statistically valid, robust form that is consistent over a 20-year business cycle. We ensure that these factors together give us more feedback and good stock selection. The idea is to select buckets and stocks with a high success rate. We create a portfolio of well-researched stocks based on historical measures and review the maximum drawdown over a historical period of 1-, 3- and 5- years. We seek to understand any downside to our portfolio and what kind of risks we are exposed to. The goal is not to use a risk model per se to optimize risk-to-return but to also create a portfolio with heuristic risk controls. When we say heuristic we are more interested at the stock level risk, factor level and the industry level exposures. The ultimate goal is to be an active portfolio manager with limited tracking error. We also want to make sure we take sufficient amount of systematic and unsystematic risk so that our definition of active shares is significant but not uncorrelated to the benchmark.
Q: How far back in time you go in capturing data to test concepts?
A : Our focus is to create a proprietary model based on our research and we use many variables to fine tune our model. Every day we test new concepts and ideas to make sure we are capturing alpha over a consistent period of time. Since 1999 we have been collecting data on a historical 10-year rolling basis from Russell 2000, our benchmark, but we also reconstruct the universe and store this data. We started gathering, on a quarterly basis, this kind of data on 2,200 companies; data on about 1,800 companies would meet our testing requirements.
Q: How do you use this data in your investment process?
A : We construct a universe of companies within the market cap range between $200 million and $3 billion. We buy U.S.-traded companies, limited partnerships, and some international companies that are listed in the United States. We currently have nearly 2,300 names in the universe. Over time, we have developed a bankruptcy score which helps us look at individual companies when we reconstruct the universe for their balance sheet stability and financial health. Finally the names of companies we would be buying is about 1,800 stocks, every quarter this is reconstituted in real time. We have the ranking scores of the model for the stocks that are put into a portfolio optimizer. For industry constraints, we buy plus or minus 3% of our industries relative to standard industry groups. At the time of purchase, we will limit our holdings to less than 1.5% in portfolio size. We generally hold between 90 and 110 names. Our idea generation results from the ranking process that we have developed and is entirely systemic. We let our models dictate the selection and we follow our models strictly. The human intervention begins at the next step. We engage in two kinds of fundamental validation - a data check for data anomalies and an actual understanding of the intuitive meaning of all this. The data check seeks to uncover data anomalies which can be created because either the data is wrong or because of something wrong with the intuition. For example, Company A spins off Company B; now Company B is part of our universe. We want to make sure that the data is intuitively correct for the right business and we are not buying stocks on a false intuition. Then there’s the other side of the fundamental validation. We don’t talk to management but we research a company through publicly available data and broker research and we attend conferences to collect information. This information is put together to help us create a composite picture for the company. Moreover, we make sure that we collect information about management review or an FDA drug announcement or any U.S. regulatory investigations that we find in news media or on the Internet. Today, we have a database of close to 2,500 different variants of this information. These validations and data collection are continual; our main purpose is to make sure that a company information we store does not have flawed data and is clean and accurate.
Q: How do you monitor and improve your model factors?
A : In 1999, we were monitoring our model factors once a month, quarter or year. Now, all the data comes in daily but factors are still monitored at a monthly level because our mandates are more long-only with an investment horizon of 12 to 18 months. We don’t change our overall model drastically unless something has dramatically changed. What we find is that a factor goes in and out of fad or based on responses from the market. For instance, in 2013 in the small cap world, there were times when price and earnings momentum was the focus but valuation was ignored. We didn’t change our model then because a model remains effective if four out of six or five out of seven factors are still working. From the 2008-crisis, we certainly learned that markets were volatile and factors changed wildly and quickly relative to the data. So we created a pseudo model that we call a “dynamic model” which is short-term focused on performance of the factors for three months. The model looked at the last three months and showed us five of the best performing factors that we don’t use. And it also provided a way to measure how these five factors collectively contribute towards something that we are missing. We take that best factor model, screen our buy-universe for this new model, and see if it is helping us without actually altering our philosophy of fundamentals on a stock. And what we found is that as long as we keep buying the names that are in our original model and not ignoring the best factor overlay model, we tend to create less turnover and more stability in our portfolios and have not missed anything in the short period of time. We tested our “dynamic model” for three years and started implementing it around 2011. That was our general idea of improvement. However, our process is geared towards identifying companies that are consistent performers rather than home runs. But we do need six months of data before we even invest in a company. So, the success of our growth strategy is mainly due to our focus on quality names with seven out of ten growth momentum factors. The times we get really hurt are during market “bubbles,” when we lag the market averages. We are always researching our factors, looking to improve our factor selection and looking at academic research to see how it will enhance our models. We don’t invest in mature companies or in companies that are just entering the growth phase. We generally invest in companies that are in the second or third phase of growth in a five-phase growth cycle.
Q: What other factors involve your research process?
A : We begin looking at companies after six months of publicly traded history and after the company has released at least four quarter of earnings data. This history helps us in gathering how the company performs as a public entity and how analysts are viewing the company’s prospects. Also, the history allows us to gather information on cash flow and earnings surprise, if any.
Q: What is your buy and sell discipline?
A : Our buy ideas come from the ranking process and are driven by the concept of our portfolio optimization. We would look to sell companies when the company begins to fall in our ranking process. We have a strict risk discipline of staying within the small cap range. At any given time there will be 80% of stock in the small cap world. There are times when companies get investigated by the government or regulators and something goes wrong. During those times, we make a call on whether we want to stay in this company or get out. We reach out to the markets or the specialists in that company’s industry to find out more so that we can determine if we are making the right decisions.
Q: How do you construct your portfolio?
A : Currently our small cap growth portfolio has close to 110 names. It is a diversified portfolio with exposures to all 10 sectors, but we constrain it at the industry level so that we have the flexibility to go underweight or overweight by 3%. And we are prepared to overweight individual names by 2.5% relative to the benchmark, the Russell 2000 Growth Index. We consider metrics such as estimate revisions, earnings dispersion, earnings surprise, net margin change, cash flow metrics, and price momentum factors. In addition, individual rank scores are created as an overall alpha score for each of the security in the universe of 2,300 names. We focus on the top 400 names that we run through the optimizer. Out of these 400 names we create a portfolio based on the cost of turnover, systematic and unsystematic risks, and the industry and individual name constraints, and we add names with the best ranking and drop stocks that are poorly placed on our ranking process.
Q: How do you define and control risk?
A : Our risk is defined hard constraints and we measure risks at the individual security level and industry level and also at the exposure levels that we track factors. So we talk about an individual security’s risk broken down into systematic and unsystematic risk. Next we break down the risk from the perspective of factor exposures or growth exposures like earnings growth, profit of change growth. These exposures are measured systematically relative to our benchmark. Then we look at risk from the perspective of industry constraints plus or minus 3% hard controls so that when it reaches above 3% it will force us to sell and when if it is below 3% it forces us to buy, to some extent. The optimizer does most of our work but we also monitor our exposures on a daily, monthly, and quarterly basis. Managing risk to us means only controlling the exposures. We don’t try to manage it around third-party risk models but rely more on our heuristic empirical constraints that we have tested.
