# Should You Make Portfolio Decisions Based Upon Historical Correlations?

Market correlations explain to us the similarity in performance and behavior by different types of assets.

For investors, correlation comparisons might be between two commodities (NYSEARCA:DBC), two stocks, a stock and a commodity, or even a stock and an index. Should correlations be used to make investment decisions?

Over the past three years, the Technology Sector SPDR ETF (NYSEARCA:XLK) and Apple (NasdaqGS:AAPL) have had a 0.58 correlation. That means a 1% increase by either, would imply a 0.58% gain by the other. The more positive the correlation between assets, the more likely both are to move in the same direction.

(Audio) Listen to Ron DeLegge @ The Index Investing Show

Correlation comparisons can also be applied at the industry sector level too.

For example, the Consumer Discretionary Sector (NYSEARCA:XLY), is the most highly correlated sector to (0.82) to S&P 500 technology stocks (XLK) over the past three years. (See table below) On the other hand, utilities (NYSEARCA:XLU) and consumer staples (NYSEARCA:XLP) have had the lowest correlation to tech stocks among the nine S&P 500 sectors over that same period.

At times, correlations can be negative.

For instance, if an asset has a -1 correlation, it means that a 1% gain by another asset or market would imply a 1% loss. And a zero correlation means there’s no statistical relationship between two assets or markets.

Correlations between assets will sometimes be casual, complimentary, parallel, or reciprocal. Yet, these relationships are in constant flux each day and never remain the exact same.

During extreme market conditions, correlations that seemed to exist between two assets can suddenly vanish and correlations that didn’t seem to exist can suddenly appear. And that’s precisely why making future investment decisions based solely upon historical correlations can be misleading. Because the future is never quite the same as the past.

Follow us on Twitter @ ETFguide

You explanation of correlation is totally wrong. What you are explaining are Beta’s not correlation. Correlation of 0.58 does not mean a 1% move in one will imply a 0.58% in the other – that is the definition of beta. Correlation is just the “strength of association” not the magnitude of association.

Hi Ravi, If the market is always up 10% and a stock is always up 20%, the correlation is 1 (correlation measures direction, not magnitude). However, beta takes into account both direction and magnitude, so in the same example the beta would be 2 (the stock is up twice as much as the market).

Not at all, Ravi is right and Ronald you are still quite confused. Suppose the correlation between two returns is 0.5; then when one return is *one standard deviation above its mean* we should *on average* expect to see the other one *0.5 standard deviations above its own mean*. You have made two mistakes. First this statistic gives information about the size of moves in terms of standard deviations for each variable, not in absolute terms. Secondly, there is no “guarantee” here — it talks about moves on average. Correlations don’t suddenly appear or disappear. They don’t give 100% explanatory power for each observation. (Absent perfect correlation.) They just give information as to the average relationship over time. A couple of outlying observations don’t invalidate longstanding historical relationships. Anybody who thinks that is true is naive. Outliers are to be expected, given a large enough sample.

Great points Markov, especially about outliers and sample sizes…or as I like to say: 99% of all statistics only tell 49% of the story 100% of the time.