The stock market has made its final prediction: Joe Biden will win the presidential election

We’ve finished the last trading day in October, meaning the "presidential predictor" is officially sealed.

Our mission to help you navigate the new normal is fueled by subscribers. To enjoy unlimited access to our journalism, subscribe today.

Even when the polls were wrong in 2016, the stock market’s time-tested “presidential predictor” was right about who would win the election: The S&P 500’s 2.2% decline in the three months leading up to November four years ago signaled that the incumbent party in the White House—the Democrats—would be replaced.

In other words, the stock market predicted Donald Trump would defeat Hillary Clinton despite polls to the contrary.

Now, the same stock market indicator—which is dependent upon S&P 500 performance for the three months from August through the end of October—has finalized its prediction. With the S&P 500 down slightly (just 0.6%) over that period on the last trading day of October, the stock market’s presidential predictor, as it’s known by market analysts, is officially signaling that Joe Biden will win the election.

Though the dip is minor, the negative S&P 500 performance over those three months indicates that the incumbent party—in other words, President Trump—will be voted out of the White House and replaced with a Democrat.

The pattern has held true for almost a century, since 1928, according to Sam Stovall, the chief investment strategist of CFRA who has long tracked the S&P 500. Since 1944, negative performance by the stock index over the key time frame has correctly predicted a changeover in the presidential party 88% of the time (see chart below). Only once has it been wrong, in 1956. (Meanwhile, positive performance by the S&P 500 in that time period has indicated reelection of the party in the White House 82% of the time.)

While there is no proven explanation for why the stock market has such a high batting average when it comes to the winner of the presidential election, analysts theorize that it likely has to do with uncertainty around the future. Investors tend to sell stocks when it’s unclear how events will play out—just see earlier this year when the coronavirus plunged the world into chaos and caused a bear market. Therefore, the thinking goes, the selloff may reveal uncertainty about how a President from a new political party will shape policy in the coming years.

Even though the S&P 500’s decline is small over the relevant prediction period, it can still be right: In the 2000 election when George W. Bush defeated Al Gore (who was then the incumbent party, following Democrat Bill Clinton), the S&P 500 was down just 0.1% from August through October, according to CFRA. Of course, the 2000 election was notoriously close, with a highly contested result that was ultimately decided by the Supreme Court—something that many politicos are already expecting could happen with this year’s election, given the unprecedented nature of the pandemic and changed voting protocols.

Still, cautions Stovall, “the presidential predictor implies, but does not guarantee, a Biden victory.”

Whether the predictor will succeed in 2020, investors and voters will have to wait until at least Election Day on Nov. 3 (or likely longer) to find out. But no matter what happens next week, stocks have officially sealed their prediction.

More must-read finance coverage from Fortune:

• Chobani and PayPal are paying workers more—and rethinking capitalism
• When it comes to climate change, says Mark Carney, this financial crisis is different—and maybe better
• Ray Dalio on why Chinese capitalism is on the rise—and why American capitalism is in desperate need of a fix
• Microsoft’s cloud could be a bit foggy for the next quarter
• Coinbase launches crypto debit card in U.S. with 1% Bitcoin reward