Originally published in the Brandywine Asset Management Monthly Report.
The addition of Brandywine to a portfolio containing stocks or managed futures can both increase returns and decrease risk. This is clearly evident in the following two metrics.
First, performance: Since we launched Brandywine’s Symphony program in July 2011 it has gained +18.69% while both the BTOP 50 and Newedge managed futures indexes have fallen (-3.00% and -4.42%, respectively). Brandywine’s more aggressively managed Symphony Preferred gained +88.51%, outperforming the +49.33% earned by the S&P 500 total return index. Clearly, adding Brandywine to a portfolio that contains either managed futures or stocks would have increased returns.
Second, diversification value: Brandywine has made money when stocks or managed futures have lost. This, of course, is obvious when looking at Brandywine’s positive performance versus the losses in the managed futures indexes. But the same benefit is even more evident when comparing Brandywine to the S&P 500. For example, when the S&P 500 dropped -13.87% from July through September 2011, Brandywine’s Symphony Preferred gained +30.94%. So adding Brandywine to a portfolio that contains either managed futures or stocks would also have decreased drawdowns (risk).
Looking at our results, the diversification value of including Brandywine in an investment portfolio is obvious. This value has not gone unnoticed, as is evidenced by our growth in assets and interest from new investors. That said, some of the conventional metrics used by asset allocators not only disguise this value, but actually lead uninformed investors to reach the exact opposite conclusion.
The Abuse of Correlation (Part 2)
In our January 2013 report we discussed how correlation metrics are easily and often abused by investors. In this report, we’ll give a further example of why correlation statistics can be misleading at best, dangerous at worst, or even downright ridiculous.
Newedge publishes statistical reports on a number of investment managers, including Brandywine. One of the measures they provide is the correlation of monthly returns between Brandywine’s Symphony program and the Newedge CTA index. The correlation data is what you would expect from looking at Brandywine’s differentiating performance over the past few years. In both up and down months for the Newedge Index, Brandywine’s correlation to the index has been less than 0.1 (meaning there is no correlation of returns). However, despite Brandywine’s strong out-performance, (the result of our use of innovations such as Return Drivers and Predictive Diversification) and obvious portfolio diversification value, it is possible that some investors could exclude Brandywine from their portfolio because they consider us too correlated to the other investments they hold.
Let’s take a look at why.
The chart below shows the performance of Brandywine’s Symphony program compared with that of the Newedge managed futures index during February. Clearly, Brandywine’s consistent profits throughout the month and strongly positive return of +5.70% were highly dissimilar to the underperformance recorded by the Newedge index. This simple chart makes it clear that there’s no question that including Brandywine in the portfolio would have added significant value during the month.
But during February, Brandywine’s correlation of daily returns to the Newedge Index increased significantly, recording a value of +0.62 during the month. Sure, there were some days where the Newedge Index posted a gain and Brandywine did as well (who wouldn’t want that?). But of the nine days when the Newedge index fell, Brandywine gained in five of them. And Brandywine didn’t outperform by taking on more leverage; both Brandywine and the Newedge index posted an identical 42 basis point standard deviation of daily returns. This is exactly the kind of diversification value you want to see in a manager. Yet someone focused on the daily correlation metric would have concluded that Brandywine provided little diversification value – the correlation was too high. This is a clear example of where a reliance on correlation proves dangerous to an investor’s financial well-being.
Now let’s move from dangerous to ridiculous.
For most of the month of February, the Newedge index was showing a loss. Had the month ended that way, you would expect that the already low 0.09 correlation between Brandywine and Newedge during down months of the Newedge Index would have collapsed, or even gone negative. After all, Brandywine was up a sizable amount and the Newedge Index was showing a loss. But perversely, that’s just not so. In fact, had the month ended on February 20th, at which time the Newedge index showed a -0.5% loss and Brandywine’s Symphony program showed a +4.0% profit, the correlation of monthly returns during down months of the Newedge index would have increased from 0.09 to 0.24! Despite the stark contrast in our performances, an investor relying on the correlation metric would have concluded there was a decrease in the diversification value provided by Brandywine!
But it gets even more ridiculous. Had Brandywine performed the exact opposite on a daily basis as it had in February [had we lost -5.70% rather than gained +5.70%], our daily correlation would have measured -0.62. This would have made Brandywine more attractive to an investor focused on correlation than did our actual significant positive performance!
When it comes to correlation, sometimes it’s better to lose than to win.
The Need to Understand Return Drivers
This makes clear the serious shortcoming in using correlation to find managers that can add diversification value to a portfolio. The real problem is that correlation tells you nothing about the “true” diversification value inherent in a manager’s trading. True diversification value (as well as true risk, a topic for another report) can only be determined with an understanding of the Return Drivers powering the manager’s performance.
Because of this, Brandywine has been more open than most managers in revealing the sources of our returns. In fact, we have openly presented one of our actual trading strategies on the Web (and at the end of this narrative, we provide links to our past monthly reports* where we discussed some of our trading strategies and their primary Return Drivers). We realize that many managers, who may only have a single Return Driver underlying their trading strategies, are necessarily more opaque and unwilling to divulge their single secret. This is what forces investors to revert to using proxies such as correlation in place of understanding the actual Return Drivers. But because of the inability of correlation to provide any reasonably useful information, if the ability exists to understand the underlying Return Drivers, correlation should be discarded as a measure.