During recent years when large-cap growth funds were eschewed by many investors in favor of small- and mid-cap value, the $836 million Dreyfus Premier Alpha Growth Fund/B, (BSBFX), a large-cap growth fund, bucked the trend. What allowed this fund to consistently perform well when other large cap growth funds could not? Is there a particular strategy that works better for large-cap growth than other methodologies in that environment? How has this fund has done so well over the past three- and five-years when value was getting much of the attention and most of the assets while the average large cap growth fund was starving for alpha?
“The key to all of our portfolios is that we manage them with complete dispassionate discipline. In other words, many managers are intrigued into stories about stocks and end up buying the stock for what we believe are the wrong reasons. Our approach is entirely quantitative, using models that we have tested over long periods of time,” says portfolio manager James P. O’Shaughnessy, senior managing director and director of systematic equity investments at Bear Stearns Asst Management, the subadvisor to the fund.
According to Standard & Poor’s, the fund has earned annualized average returns of 8.73% for the five years ended January 31, versus an annualized average loss of -3.29% for its large-cap style peer group; and an annualized 20.38% return for the three years versus 15.27% for its peer group. The fund earned a 30.79% total return for the one-year period, versus a 12.80% average return for its peers. S&P has rewarded this performance by giving the fund a five-star ranking for one and five years, a three-star ranking for the three-year period, and three stars overall. We spoke to O’Shaughnessy in early February.
Overall, how much money do you manage?
Overall my team manages $7.1 billion, and in terms of the fund it is closing in on $900 million. It’s been growing dramatically–Dreyfus has been doing an outstanding job marketing it to all the various financial advisors.
On the cover of your new book [Predicting the Markets of Tomorrow, Portfolio, March 2006], you describe yourself as a contrarian or with a contrarian strategy. Why?
When you’re working with publishers you have to let tem get the titles that they want. They’re arguing it’s contrarian because it’s advocating an asset allocation scheme for a 20-year period rather than the traditional sort of people who are looking for what tends to be hot this year; it also advocates investment outside of the S&P 500 Index funds, so they took that to be contrarian.
Then you wouldn’t necessarily describe yourself that way?
I would say I have been described as contrarian in media quite a bit because we tend to have strategies that look a lot different from traditional managers. For example, the Alpha Growth Fund. Traditional growth managers usually don’t concern themselves with valuation metrics, whereas our work, because of the research done in What Works on Wall Street, [McGraw-Hill, 1998], which covers now 54 years of data, suggest that even managers managing to the growth space are much better served by having valuation carburetors, if you will, in place in the fund to avoid getting into the names that might be very, very popular, but have already reached valuation levels that we believe it would be difficult for them to continue to go on to do well. So for example in the four screens that we use to select the securities for Alpha Growth we always have a value parameter lurking somewhere, and that makes us quite a bit different from a lot of growth managers in that we will not buy ‘growth at any cost.’ We like to insist that we get into names where we’re paying a lot less for every dollar of earnings or sales or cash flow. I guess that would make us slightly contrarian to a traditional growth approach.
Maybe that’s one of the reasons why the fund has been able to do well in the past several years where value has been the king of the hill. Are there other reasons why you’ve been able to manage this fund to do well in the last 3 years, 5 years when value was getting all the assets?
Well I think that the key to all of our portfolios is we manage them with complete dispassionate discipline. In other words, many managers are intrigued into stories about stocks and end up buying the stock for what we believe are the wrong reasons. Our approach to investing is entirely quantitative, using models that we have tested over long periods of time. I believe that the key to succeeding in a variety of market environments is using models that have been tested over long periods of time and using them with a sort of an iron-like discipline that you won’t override. I believe a lot of growth managers got caught up in, I guess I’d call it “yesterday’s story” and continued to hang onto stocks that they believed in their hearts were great growth stocks. We’re completely dispassionate. That means we run the screens; two of the screens are run annually, and two are run quarterly. If a stock that we’re holding no longer qualifies, it is sold. It simply doesn’t matter how we might feel about that stock. Conversely if we hold a stock that’s done very, very well, and it continues to qualify, we sell it back to an equal weighting in the portfolio. In other words, we take some money off the table, and then clearly when the models are run, any name that meets the criteria is put into the portfolio, so I truly believe–I like to joke that we’re passionately dispassionate. We absolutely only invest according to strict guidelines that have been tested.
Everything that we do begins life as a test. We have some strategies for which we have ten-year real-time track records. All of our strategies began life as a back-test–looking at how the strategy performed in a variety of stock market environments. We have access to metrics and statistics that more conventional managers simply can’t know about their own approach, for example, the Alpha Growth strategy tested from 1985 forward. Why 1985 and not 1950? Well because some of the strategies that we use in Alpha Growth use forward-looking estimates; prior to 1985 there really weren’t very good consensus forecasts available. So, we will test every strategy, to see: “OK what was the worst case for this strategy? What was the biggest peak-to-drop decline? How often does the strategy meet its benchmarks? By what magnitude? How risky is the strategy? What’s its upside capture and downside capture?” What we’re always looking for, of course, are strategies that have very robust numbers around all those things. We’re looking for strategies that basically put the odds in our favor, with good batting averages against benchmark, with reasonable risk characteristics, and peak-to-trough declines that aren’t heartbreaking. Essentially what we’re trying to find is the sweet spot.
What we also do with Alpha Growth is we have four separate strategies that populate the portfolio. These strategies reflect this not only because on a standalone basis they were excellent performers, but also because they have mild correlations with one another. In other words, [when] strategy A is on fire, strategy B might be coasting. Also, vice versa. By pushing together four separate strategies where the correlations were relatively low, what we’re able to do is get overall performance from the portfolio so that we’ve always got one of the strategies working.
Can you describe the models and how you use them?
Absolutely. We begin life with a universe of approximately 1600 companies and then we have four separate strategies that we use to select securities. The first strategy is we start with 1600 companies, but from these 1600 we’re going to limit [the portfolio] to the largest 300 stocks by market capitalization. The next screen used to make people chuckle, and wonder why we ever would do it in the bubble of the late ’90s and 2000, but that is: we require a price-to-earnings ratio. In other words, there have got to be earnings at the company for us to be willing to put it into the portfolio. They chuckled in the bubble–they don’t chuckle about that one now!
The next one is the best growth characteristic factor that we were able to find on a standalone basis, and that is price appreciation or relative strength. We limit [the portfolio] to the 100 companies that have P/E ratios, with the best 12-month price appreciation. Essentially these are the companies that have done the best over the prior 12-month period. We next put a price-to-sales governor in place: we will exclude any company that has a price-to-sales ratio exceeding 10. We do this because price-to-sales emerged as the single best value characteristic to look at in terms of value in the stock. The higher the price-to-sales ratio going forward, at least from the 54-years of data that we have, the lower the return to investors. So we want to exclude what we call “the toxic stocks,” those that are priced to perfection and beyond, and we say the most we’ll pay for $1 of sales is $10. Then we use even more relative price appreciation: relative strength or price appreciation married to reasonable valuation metrics emerged as one of the best ways to get to growth stocks that were poised to continue to do well on a going forward basis. So the final screen for the first model is that we take the 25 names that have the best 6-month price appreciation into the portfolio.
So, biggest 300, got to have earnings, got to be in the top 100 by one-year price appreciation; got to be reasonably priced in terms of what we have to pay for every dollar of sales; and we buy the 25 names that have the best 6-month price appreciation. You might have seen the book The Wisdom of Crowds [by James Surowiecki, Doubleday 2004]? We’ve found that’s sort of [Benjamin] Graham’s idea of Mr. Market. Price appreciation tends to distill down what Mr. Market thinks about a stock down to that one thing, “How well has it done? What does the market in general think about the prospects for this stock?” What we’ve found is that price appreciation works extremely well to get to the names that should continue to do well on a going-forward basis.
The second model requires all the same things that the first model requires, except here we buy the 25 names that have the highest consensus forecasted growth rate and earnings for the coming five-year period. Here we’re bringing market opinion into play. What we find is that these two models have relatively low correlation with one another. In other words, these are names that the analysts, on a consensus basis, think are going to do the best on an earnings compound growth rate, for the coming five-year period. Here, what we’re trying to do is bring in opinion to mix with the price appreciation screen. This gives us another 25 names. I should note that if we have similar names in model I and model II, we only buy the stock once. We eliminate any duplications.
The next two [models] are actually 10-stock screens. These were brought in because they have very, very different characteristics than the first two dominant models, but they get us some nice diversification in terms of the profile of the security as well as that low correlation with the first two models. The third model looks for a classic growth [pattern]. Limiting itself to the largest 400 companies by market capitalization, it then puts that same price-to-sales value governor into place, but then we require that the stock have historical one- and five-year earnings growth rates exceeding 20%; forecasted earnings growth rates for the coming one- and five-year periods also exceeding 20%; and then we buy the 10 names that score the highest on a price stability index that we maintain. Essentially, if you get a one on a price stability index you have no price stability. If you get a 100 on that index you are one of the most stable stock prices. We buy the 10 names that meet the other criteria that have the highest price stability score. A classic large-cap well known growth company where we still have good valuations in place; it’s demonstrated superior earnings growth over the prior one- and five; panelists think it’s going to continue to be a superior earnings grower over the next one-and five-year period; and its price is very, very stable in the way the price moves.
The final model actually uses the S&P 500 as its universe. We value every stock in the S&P 500 according to seven separate valuation and price metrics. This is what’s known as a summation model, whereby we’re not funneling like we do on the first three screens, but we’re giving each stock in the S&P an ordinal rank, meaning we rank on working capital, and price appreciation, and price-to-sales ratio. To use price-to sales as our example, if a stock in the S&P has the highest price-to-sales ratio in the index it receives an ordinal rank of one. If it has the lowest, it gets a 500. Same with price appreciation–the best price appreciation gets a 500; the worst price appreciation gets a one. We then sum up all of these ordinal ranks and we buy the 10 names on the list–that don’t already occur in the first three models–with the highest ordinal rank. So, unlike the funnel system you could, conceivably, get into the [portfolio] if you might have a high valuation but you score the best on all of the other characteristics–working capital, profit margins, things like that. This doesn’t exclude based on funnel systems but it does a summation model, which we’ve been doing a lot of work with recently and found that they can lead to outstanding performance.
The other part of this model is that it leads to a very different [stock] than the first three, so that gets us into a situation where we are having in the portfolio a variety of names with different characteristics, and I think that speaks to the reason why the last three years look so good, because we can get into names that are definitely counter-intuitive, and by doing so, we’re there in the market as that name is doing well. Two of these models are re-run on a quarterly basis, the second model and the third one–the one with earnings forecasts, and the one with price stability. Every quarter we re-run those models and our rebalancing procedure is our way of saying “This is the ultimate sell discipline.” When we re-screen the model, any security that no longer qualifies gets sold. So, our opinions don’t matter, our hopes, dreams, aspirations–it really doesn’t matter; if it no longer meets criteria, it’s sold out of the portfolio. If it continues to meet, it’s, again, sold down to an equal weight, and then obviously at that time we add the new names.
People ask: “Well, what happens if a sector is really hot and starts to cool off?” Well, on the rebalance we move out of that particular sector, and what we’ve found is that these work very, very nicely with each other on an ongoing basis. [In addition] to our calendar-based rebalance, we also have a series of red flags in place that would allow us to sell a security out of the portfolio prior to the formal rebalance. The first is, if a company fails to certify their numbers or restates their numbers such that when they restated them they would not have qualified at the point of purchase, we remove the company from the portfolio. Number two: if the federal government alleges fraud against the company–generally a bad thing–we will remove the company from the portfolio. The third basically happens well ahead of one and two, and that is if a company’s price declines from our point of purchase by 50%, we remove it. That’s an empirically derived number; [we] tested a variety sell-stops like 10%, 20%, etc., and I found it interesting that I’d often thought 10%-20% would be a good sell-stop; basically you remove all of the alpha of the portfolio if you have stops that close in, meaning the market’s a bit more volatile than a lot of us intuitively believe it is.
So you have to endure that bounce?
You have to endure the pain of that shorter-term volatility. However, once it reaches that 50% it’s like just falling off a cliff, basically. Then you’ve got relative strength working the other way, and so you want to get out of that name–it’s a bit like saying if we were unlucky enough to own Enron, we’d rather sell it at $40 than at $4, or four cents. You remember during Enron’s decline there were a lot of people [who], when it declined from $80 to $40, were saying, “Oh, this is a great stock, you’ve got to get into it.” We would have said, “Nope, we’re getting out of it, because it’s declined by 50%.”
If a company is taken over during our ownership of it, and the acquiring company does not meet our criteria, we will sell the security out of the portfolio, and invest the proceeds in the remainder of the portfolio. So we will allow a takeover if it’s a company that fits our criteria, we would take the stock of that new company.
And those are your two basic ways of rebalancing?
Also if a company goes into bankruptcy, it is also sold out of the portfolio. That’s really more procedural than anything else because, typically speaking, we haven’t had a company that ends up going into bankruptcy that hasn’t already declined.
It wouldn’t be so abrupt, normally?
That’s right, unless they fooled the market–company’s going along and doing well and then suddenly, “Oh by the way, we’re going bankrupt.” I guess, conceptually, that could happen, but it hasn’t happened yet, let’s put it that way.