Enhanced Investment Technologies Inc., known as INTECH, has become well known for using mathematical processes to optimize its portfolios. The firm — drawing off the "stochastic CAPM" model created in 1982 in part by its chief investment officer, Robert Fernholz — uses a proprietary, mathematical investment process that takes into account the fact that the returns of individual stocks in the market are random. In 1997, Messrs. Fernholz and Garvy extended the stochastic CAPM work by introducing a process called diversity-weighted indexing. This process questions the wisdom of the traditional cap-weighted index.
By using this approach, INTECH has consistently outperformed its benchmarks and has quietly become one of the most respected — and fastest growing — equity management firms in the institutional investment world. The firm recently launched a global equity strategy pegged to the Morgan Stanley Capital International World index that uses the same mathematical approach. The firm's assets under management nearly tripled to $45 billion at the end of 2005 from $17.5 billion at the end of 2003. When Janus Capital Group Inc. acquired a majority stake in the firm in 2002, INTECH had only $6 billion in assets.
Explain, in layman's terms, diversity-weighted indexing. The difference between what we do and what the typical manager does is that we do not do fundamental research to predict the direction of the movement of stocks. What we do is take the movement that is created by those who are attempting to predict the future direction of stocks, and apply to that volatility rigorous, mathematical and statistical processes that will allow us to extract an excess return from the natural volatility itself.
Essentially, if you think about the market in a general sense, you have a market return, and then you have some stocks that have a higher return than the market, and some stocks that have a lower return. What we're doing, in a very disciplined manner, is sell small amounts of the stocks that are rising relative to their optimal (benchmark) weight and invest those funds in a very disciplined way among the securities that are declining relative to that benchmark.
So when you step back from that, that's Investing 101; it's buy 'em low, and sell 'em high. But we're doing it with a very sophisticated mathematical model. These stock price movements tend to be random. You'll get some that move up, and some that move down, and then they tend to move toward that mean. This is not dependent upon mean-reversion, but that's the fundamental idea of what we're doing. To put it simply, we're providing liquidity to the market. But overall, we're controlling the risk of the portfolio, and that's the key thing that differentiates us.
What kind of tweaks, if any, have you made to this model since its creation? There are continuous tweaks. We're constantly attempting to find another way to improve the efficiency and effectiveness of what we do. For example, in the summer of 2001, we went from a quarterly optimization … to a partial weekly reoptimization. That took the size of the trades down, thereby reducing implementation costs and increasing capacity very significantly. In 2004, we introduced a revision of the covariance estimation process — how these stocks relate to each other — and we went from a one-year historical look back to a four-year look back, because of ways we found to more effectively use that information. These sorts of things are quite modest in their impact typically, they amount to a few basis points. But if you can do five or 10 of those over five to 10 years, you cumulatively end up with a far more efficient process.
Do anomalous periods, such as when a few stocks or sectors are driving the returns of the market, affect the performance of your model? Absolutely. There's no question of that, because our strategy is an active strategy. We are not holding the weights of the stocks equal to the weights of the stocks in the index. We're altering those weights to take advantage of our correlation and volatility analysis. So when a sector of the markets dominates — such as larger-cap growth stocks in the late 1990s; the oil stocks now; the conglomerates back in the 1960s; etc., whatever happens to be driving the market at a particular time — those areas create tracking error. A key part of what we do is to control that tracking error. Just as essential to us as capturing the excess return from the volatility is to control the risk of the portfolio relative to the benchmark by controlling those anomalous situations.
As quant managers, do you believe in efficient market hypothesis? What we do is make a distinction between an efficient market and an efficient portfolio. In other words the market, as described, say, by the floor of the New York Stock Exchange, may be efficient in an Adam Smith Wealth of Nations kind of context in that there are many buyers and sellers, there are readily available prices and information, transaction costs are low, etc. When you get that type of situation, prices will reach equilibrium very quickly. That's what we mean by an efficient market. But an efficient portfolio, as described by Harry Markowitz, must take in to consideration the relationships stocks have with each other — the covariances — and cap-weighted indexes don't do that. So, therefore, what we say is these cap-weighted indexes cannot be efficient portfolios. We're not saying we have the most efficient portfolio, but by applying sophisticated mathematical and statistical processes, we can create a portfolio that's more efficient than those cap-weighted indexes, and we've been doing it now for over 18 years.
Where do you envision the firm being five years from now? When we designed the model for INTECH, the intent … was to be able to manage extraordinarily large pools of capital effectively and efficiently. When you look at our portfolios, they typically will have hundreds of securities in them. And those positions will be dominated by the large-cap stocks because they are the largest in the market. We have built into our model extraordinary capacity. So I would be very surprised if, in five years, we were not substantially above $100 billion in domestic equities.
It seems that quantitative money management has made a comeback since 2000. Would you agree? Absolutely. I think there are two very specific reasons for that. Prior to 2000, the stock market averaged 17% in returns per year. Plan sponsors had no difficulty watching their assets grow substantially, and typically much faster then their liabilities. Beginning in 2000, however, we had a massive correction. Investors were dramatically reminded of how sharp this two-edged sword of investing can be. Risk management came back in to vogue. That brought quantitative money management back very strongly. Anyone who had a 401(k) and suffered the kind of pain they did became conscious of the need for risk management. The second reason is that the type of processes INTECH follows have been greatly enhanced over the past decade by the tremendous advances in computers and the availability of supercomputing capacity for research and the ability to apply advances in statistical and mathematical theory to real-world problems.