Can Two Negatively Correlated Investments Both be Positive YTD?

With stocks back at all-time highs, one might think managed futures programs – you know, the poster child of performing during stock market crisis periods , would be sucking wind. After all – managed futures (SG CTA Index) posted gains of +7% back when stocks took a dive (falling -10%) in January-February, so you might think they would be down some ugly percentage while stocks have rallied back over 19.41%.

But something funny happened on the way to turning in their asset class scorecard – managed futures found return drivers in metals, bonds, and even stocks themselves – to actually move to positively correlated with stocks right now – just five months after being negatively correlated to them.  You can see the Attain Short Term Fund in the table up +13.75% on the year, at the same time stocks are up 7.75% on the year – all while having an all time correlation of -0.36 and seven-month correlation of -0.89.

(Disclaimer: Past performance is not necessarily indicative of future results)
Stocks = S&P 500

What… How can two negatively correlated investments both by in positive territory for the year?

Let’s first back track a minute and define correlation.  Here’s a simple review from an old newsletter of ours:

You can think of correlation as the numeric  connection  between two events. The number of people wearing shorts in Chicago, for example, is  positively  correlated with how warm the temperature is, while the number of people wearing gloves is  negatively correlated with how warm the temperature is. The warmer it is, the less people wear gloves and the more people wear shorts. While we can easily see that connection in our minds –  seeing  such connections between investment returns can often prove difficult.

More specifically, correlation is a statistical figure with values which range between -1.00 and +1.00, meant to show how inter-related two sets of data are (in the case of investments, we are usually looking at the monthly percentage returns). If they have a correlation of 1.00, they are perfectly correlated, meaning when one market rises 3%, the other will do the exact same, and when one loses -2%, so will the other. If they are at -1.00, they are exactly opposite; with one making the exact opposite amount the other loses each month, and vice versa.

While not statistically correct, a good mental shortcut you can use is to think of the correlation statistic as a sort of percentage of similar readings. So, a reading of 0.50 roughly means that 50% of the data set moved together, -0.80 meaning that the data set moved in opposite directions about 80% of the time. 

Where that shortcut doesn’t work, and where correlation gets confusing is at values between roughly -0.5 and +0.50. That is the range of NON correlation, where the data has no discernable pattern, sometimes moving in tandem, sometimes moving opposite. Problem is, it is hard to visualize the absence of something – and most investors equate non-correlation with negative correlation as a mental shortcut.

We find it helpful when thinking about correlation to imagine it not as an absolute value defining what will happen each month, but rather as an average of what happens month to month. We all tend to think a long term correlation reading of around 0.00 means a bunch of shorter term correlations of around zero – but it can just as easily mean a bunch of high positive correlations of +0.80 or higher averaged with a bunch of negatively correlated values -0.80 or lower.

Just look again at the short term fund – which posted gains of +8.73% in 2014 when stocks were up +13.53%; posted a small loss of -1.85% last year when stocks were up 1.34%, and now finds itself up 13.75% on the year when stocks are at all-time highs. That is what non-correlation looks like, a ragged mixture of positive and negative correlation depending on the time frame you’re looking at, all averaging together to result in non correlation.

To see more performance and how it works, click here to download the Short Term fund factsheet .