The Benefit of “Predictive Diversification”

Originally published in the Brandywine Asset Management Monthly Report.

Throughout the summer our monthly reports focused on the “perfect” drawdown Brandywine incurred from April through July. We called it perfect because, as we described in those reports, it was right in line with what we expected based on our research and past trading. It was our confidence in our research methodology and the statistical validity of our past performance that led us to write in early July that “now may be an excellent time to invest with Brandywine.”

Another factor that prompted us to recommend new or additional investments was the increase in Brandywine’s market exposure at the end of July. Brandywine employs dozens of fundamentally-based trading strategies, each based on a unique return driver, to trade across more than 150 global financial and commodity markets. When market opportunities arise, an increasing number of those return drivers signal trading opportunities. As a result, our overall portfolio exposure across all strategies and markets is often highest during or immediately prior to our best-performing periods. This is exactly what occurred with Brandywine at the end of July.

During the drawdown (-7.55% over four months), Brandywine’s market exposure, as measured by our margin-to-equity ratio, averaged a below-average 7.81%. We pointed this out in our July report and then wrapped up the discussion by stating that on the last day of July our market exposure (measured by our margin-to-equity ratio) “increased to its highest level in more than four months” and that “this indicates that Brandywine’s Symphony Program is confident about near-term profit opportunities.” July 31 marked the low point in Brandywine’s performance for the year. Since then Brandywine’s Symphony Program has gained +7.10% and our aggressively-traded Brandywine Symphony Preferred Fund gained +30.45%. This rally was accompanied by an average margin-to-equity ratio of 9.59%.

Not every performance drawdown behaves as well as the one we experienced this past summer. And while the odds favored a performance rally at the end of July, there was of course still the possibility that the drawdown could have extended for a longer duration and greater magnitude. All we (Brandywine and our clients) can do is play the percentages, which indicated an approaching end to the drawdown.

Perhaps the best question to ask is, “Why was Brandywine so confident in the percentages?” The answer is due to our use of “Predictive Diversification.”

Predictive Diversification

Predictive Diversification was developed by Brandywine more than twenty years ago. We were systematizing our return driver based trading strategies and sought a process for allocating to each of them in our portfolio. At that time, and remaining to a large extent today, the conventional approach employed by portfolio managers to allocate across the constituents of their portfolios was to use some variation of “mean-variance” modeling such as modern portfolio theory. In this approach, which contributed to Harry Markowitz receiving the Nobel Prize in Economics in 1990, an “efficient frontier” is identified where various allocations result in the best returns for any given level of risk. Unfortunately, this model, although being academically elegant and quite precise, produces results that are simply wrong. Virtually everyone who uses this approach modifies either its inputs or its outputs (such as by using constraints) in order to avoid unrealistic results. They realize that the optimal results derived from the model are unlikely to persist in the future.

Brandywine addressed this issue in the same way we approach all research problems. By starting over. If the output of any of our research projects produces knowingly incorrect results, there is no adequate tweak that can fix it. The project must be redesigned from the start.

In applying this approach to the portfolio diversification problem, we came to realize that modern portfolio theory was the right answer to the wrong question. The question should not have been “How do I solve for the best risk-adjusted return of a portfolio?” The question should have been (and still is!), “How should I allocate within my portfolio in order to achieve the most predictable returns?” (After all, if future performance is knowingly unlikely to match the past, nothing else matters.) It was that question that led to the development of Brandywine’s Predictive Diversification portfolio allocation model. The use of Predictive Diversification, in combination with Brandywine’s return driver based trading strategies, results in portfolio allocations that are significantly different than those produced using conventional portfolio diversification models. But most importantly, Brandywine’s model results in a portfolio that gives us more confidence (than if we had employed conventional portfolio allocation methods) that past performance will carry into the future. After all, that is what we solved for.

If you’re interested in learning more about the unique research philosophy and methodologies that have contributed to Brandywine’s differentiating performance, please contact Rob Proctor or Mike Dever to set up a presentation.

Share this post:
  • Facebook
  • Twitter
  • email