Correlation is something that is often cited by market analysts — when X happens, it causes Y to happen in the stock market. But in his latest piece for Forbes, Kenneth Fisher says simple statistical analysis shows that many of the alleged causes of market movements are more myth than reality.

“Almost every day you can find in media commentary that XYZ is causing stocks to fall (or rise),” Fisher writes. “Such definitive statements are common — but what’s almost always missing is statistical proof. And if you cannot prove, statistically, two things are linked, you don’t have much basis for believing Event X causes Outcome Y, no matter how strongly most believe it. And if you can’t prove two things are statistically linked, you shouldn’t make market bets based on them.”

One example, Fisher says, involves oil prices and stock prices. One commonly held belief is that rising oil prices are a strain on consumers, and thus on the economy and stock market. But Fisher says that it turns out that since 1973, oil prices and stock prices have a correlation of -0.003, which he says is “much too low to matter at all. Many would round that to an even zero. So as compelling as headlines can be about high oil being a negative for stocks, over longer periods, there’s no statistical evidence the price of one has any meaningful relationship to the price of the other. Over short spurts, sure, they can be strongly negatively correlated. But so too can they be strongly positively correlated for periods. Over longer periods — the relationship utterly breaks down.”

Fisher looks at other similar relationships, including stock prices vs. the U.S. dollar, and explains how you can do simple correlation tests on your own using a spreadsheet program. “Because very frequently, headlines proclaim relationships between factors that aren’t statistically provable,” he says. “You don’t want to make market bets on relationships that don’t actually exist — not when it’s so easy to test.”