But we never change the portfolio instantly, just on the basis of the data from the last three months, because in many cases a two-month period doesn’t signify a new environment. Sometimes it is an adjustment or a shock, and sometimes it is an actual regime change.
So the challenge is to react quickly enough to capture the regime change and slowly enough to avoid getting whipsawed. If your strategy is very responsive, you’d be continuously chasing your tail. That is why it is crucial to have a disciplined process, which keeps human emotion out of it.
We may test shorter than 36-month windows because we live a world where people react faster, but we would do that test every five or six months to see whether we should change the model. Most of the time, we don’t see a big gain from that change.
The day-to-day process of managing the portfolio is heavily systematized and accounts for about 20% to 30% of our effort. The remaining 60% or 70% are devoted to looking at new factors to test and incorporate in our model, if proven to have an effect.
From a research standpoint, that’s where we spend the bulk of our effort as opposed to going out, visiting companies, and doing the fundamental, data digging type of research. In our universe, especially given Regulation FD, there is no superior access to information. But we can aim to process that information better than everybody else.
Q: What factors have you found changing in the past two or three years?
A: During the past two-three years, coming up with a quantitative way to measure leverage really helped us. Most people think of leverage as the debt-to-equity ratio, but this ratio actually hides a lot of leverage, such as the leverage from leases. We measure leverage as the sensitivity of a company’s stock price to changes in junk bond returns. It is a company’s beta with respect to junk bonds, which means that companies with high beta to junk bonds are highly leveraged, and vice versa.
In the current environment, the companies with high beta sensitivity to junk bonds did really well. That was a significant long position in our portfolio in the recent environment of easy access to credit and low interest rates. But you have to find a way to measure, quantify, and incorporate the measure into your process.
Five years ago, we added insider trading data to our model because there is very systematic data from the SEC on insiders buying and selling. Through testing, we found that insider selling wasn’t that informative, but insider buying was very informative. Most insiders own a lot of company stock and tend not to buy more stock unless something attractive is going to happen. The lack of insider activity tends to be a very negative signal, because most insiders are either buying or selling stocks. If you’re an insider, and you know something bad is going to happen, you’re not going to buy the stock, but you can’t sell it because that’s illegal.
Q: What are the main considerations when constructing the portfolio?
A: A major consideration is whether the positions are expected to outperform their respective indexes after we account for the expected trading cost. The portfolio is similar to the S&P 500 in terms of average market capitalization and sector exposure, but differs in terms of holding as an actively managed long/short portfolio.
Right now we have about 120 stocks but the number varies depending on the volatility of the market and because we’re targeting the specific risk level. For example, during the tech bubble, we had about 200 stocks in the portfolio, while a year ago, when volatility was low, we could afford a relatively concentrated portfolio of 85 stocks. So we don’t target a specific the number of names. Rather, the number of names is a function of the volatility.
Q: Why have you chosen to target a specific volatility level and not a risk-adjusted return?
A: From our perspective, when somebody buys this fund, he or she buys exposure to the large cap U.S. market and the alpha that we generate on top of that. We target similar volatility to the S&P 500 to provide the large-cap exposure or behavior that investors are looking for. If we allow twice the volatility, from an asset allocation standpoint, investors will end up with more volatility in their portfolio than they desire.
Q: What’s your buy and sell discipline and the catalysts that you look for?
A: The main catalyst is the ranking process. On a daily basis, we collect all the data on all the companies in the universe, and we score each stock from 1 to 1,000 based on onemonth expected return. Then we look at the portfolio to see whether it has a higher score than the index and we have a range and a target for that. If it is lower than the target, we rebalance the portfolio taking transactions costs into account.
When we do rebalancing, we don’t just buy one name and sell one name. We go through an optimization process to make sure that the portfolio keeps its volatility and sector exposure similar to the index, and then we buy and sell a host of names to raise the average ranking of the portfolio. So portfolio construction is not a name-by-name process, but focuses on the entire portfolio because that’s what the client’s holding.
Q: Trading costs have come down substantially in the last five or ten years. Why is it important for you to focus on the cost element?
A: In the large-cap universe, there are stocks that are very cheap to trade, and there are stocks that are very expensive to trade. When a stock is mispriced, or 10% below your estimate of fair value, it has 10% expected return. If it costs you 2%, including market impact, to get in, and 2% to get out, you’re only left with 6% expected return. That means that you may be better off with a liquid stock that has expected return of 7%, and costs of 0.5% to get in and 0.25% to get out. We’re always looking at that tradeoff and the ability to get in and out in a meaningful way. Overall, we look at the net cost of ownership of each stock.
Q: What is your view on risk control? |
