The stock market’s behavior took a distinct turn in February, as it cast off the bubble wrap that apparently enveloped it throughout 2017. Indeed, after posting only eight daily returns exceeding +/-1% last year, the S&P 500 experienced 12 such days in February, and nine in March (the average over the past decade has been 65 per year).

With the return of volatility, investors, journalists, and fund rating services immediately put alternative funds under the microscope to determine if they lived up to expectations. Many quickly concluded — incorrectly — that alternative funds failed. This was a result of confusing low correlation with negative correlation, and by conditioning on risk statistics calculated with monthly returns.

Let’s start with correlation. Over the last 15 years ending in December, both the Bloomberg Barclays U.S. Aggregate Bond Index and the Credit Suisse Managed Futures Index have had correlations to U.S. stocks of essentially zero (0.00 and 0.07, respectively, to be exact).

Further, the correlation between these two indices has only been 0.24. Both bonds and managed futures strategies have been great diversifiers for stock portfolios and relative to each other. And in case you are wondering, their return profiles have been quite similar as well, with the Aggregate Index annualizing at 4.15%, and the Managed Futures Index returning 3.82%.

While these assets have exhibited no meaningful correlation to stocks over time, they have not been negatively correlated to stocks. A negatively correlated asset would be expected to rise as stocks fall.

In addition, correlations between stocks and bonds, and stocks and managed futures, are unstable. Bonds and managed futures have had periods of both high positive correlations and high negative correlations with stocks, but you can’t really plan on one of those regimes.

You must expect correlations to be in line with the long-term data. You should expect correlations to be close to zero over a meaningful holding period.

To expect a negative correlation between stocks and managed futures strategies over a period of a few days or weeks, following a period when trend-following strategies were riding the positive trend in stocks, is to completely misunderstand trend following. Trend-following managed futures strategies performed exactly as could have been expected (for the record, we don’t manage trend-following strategies.)

What about the data frequency used to calculate risk statistics?

The industry standard is to calculate risk statistics using monthly returns. While this gives a realistic profile of what to expect over time, risk statistics calculated on monthly data can provide a false sense of security at a tie when investors are obsessed with ever shorter time periods.

Let’s look at the S&P 500. How volatile have stocks been over the last 15 years? A simple question, but the answer is, it depends.

Using monthly returns, the annualized standard deviation of stocks has been 13.3. Using daily returns, though, it has been about 20. Stocks have exhibited volatility that is more than 50% higher than one would expect based on monthly returns.

Similarly, correlations and betas for many alternative strategies are low using monthly returns, but often, as is the case for some funds, returns can and will be perfectly positively correlated, with beta 1.0 performance on any given day.

That doesn’t mean that they are failing to provide the promised diversification, it means that the timeframe over which they are being judged is off. When we train ourselves on monthly data, and then fixate on daily performance, we assure ourselves of being disappointed.

Statistics are important tools in portfolio construction, but to be of value, they must be properly understood. Further, the frequency of data that goes into our calculations can have a dramatic effect on the result, and with that, expectations. Keep this in mind, and pay less attention to daily performance, and you may make better investment decisions.

Cliff Stanton, CFA, is Co-Chief Investment Officer of 361 Capital.