The Abuse of Correlation

Originally published in the Brandywine Asset Management Monthly Report.

Throughout the years, we have observed an abuse of the correlation statistic in selecting managers. People seem to forget that their goal is to find managers who, when combined with each other, will increase overall portfolio returns and decrease risk (especially event risk). They are looking at correlation in order to help them find such managers, since non-correlation may be an indication of this diversification value. But instead of focusing on the goal, people have become increasingly focused on the correlation metric, which often results in them missing their goal, which is (we’re repeating ourselves here) to find managers who will increase portfolio returns and decrease risk.

Brandywine’s Symphony program has a negative –0.06 correlation to the S&P 500, a 0.07 (non) correlation to the AlphaMetrix managed futures index and a low 0.34 correlation to the Barclay CTA index. Despite that, we occasionally hear analysts report to us that they find Brandywine Symphony’s performance to be “more correlated to trend followers” than they would expect, based on the fact that our portfolio is dominated by fundamentally-based, non-trend-following trading strategies (100% systematically applied). And that may be true. We don’t specifically structure our portfolio, as do some CTAs, to be un-or-negatively correlated. We just want to make money as consistently as possible with (this is the most important condition) a high probability that future performance will match past performance. Although the majority of our portfolio is fundamentally-based, we do have a 20% exposure to trend following. When trends occur, we want to be on the right side of them, and that does boost our correlation to trend followers in those trending periods (a GOOD thing!). But what really matters is not our correlation, but our value in achieving THE GOAL, which is to “increase portfolio returns and decrease risk.” Towards that goal we succeed completely.

One indication of this is that during Brandywine’s test period 1999 through June 2011, the BTOP 50 managed futures index suffered 18 drawdowns averaging -4.74%. During those periods Brandywine’s Symphony program GAINED +0.74% and our correlation with the index during those periods was a NEGATIVE 0.06. Furthermore, in keeping with the condition that past performance must be as predictive as possible of future performance, this characteristic holds up in actual trading. Since the launch of Brandywine’s Symphony program in July 2011, the BTOP 50 suffered a -5.21% drawdown. During that same drawdown period Brandywine’s Symphony program gained +5.74%.

So why is our correlation positive overall and should it be viewed as problematic?  Well, let’s ask a simple question. . . When the BTOP 50 is up, what would you prefer if you were invested with Brandywine’s Symphony?  Would you prefer we lose money to keep our correlation negative or would you prefer we make money, at the risk that someone misusing the correlation statistic might simply view an unparsed correlation number and think that embedded in the statistic was useful information?  The only reason (the “problem” if you will) that inflates our correlation statistic, is that we make money when the trend followers make money, even though most of what we do is driven by non-trend following trading strategies.

We would contend that as long as we make money when others are losing money (which we do) and are thus negatively correlated when trend-followers are losing money (which we are) then one should treat with great caution any statistic that would lead one to allocate less to Brandywine simply because we make money when others are also making money!

One way to see our value is to simply add Brandywine’s Symphony program to a portfolio and see the effect. We encourage you to present us with your portfolio. We will run a simple analysis that shows how that portfolio would perform with various allocations made to Brandywine’s Symphony. Because of the unique return drivers incorporated in Brandywine’s Symphony program, we are confident that the inclusion of Brandywine will both increase returns and decrease risk in your portfolio.

We look forward to talking with you soon.

Share this post:
  • Facebook
  • Twitter
  • email