With both the Democratic and Republican conventions in the rear view mirror and the official nominees set in stone, our attention as voters–and American citizens–turns to the November election. But with the opportunity to cast our vote still being over two months away, many have put intense work into analyzing and interpreting polls both national and local to determine who our next president will be. Unfortunately, depending on when those polls are taken, they might not carry as much weight as you may think.

Shortly after the Democratic National Convention in late July, many news and polling outlets boomed with news that favored Hillary Clinton strongly over Donald Trump, some citing leads as large as 10 points. However the Clinton camp should not have necessarily been pleased with these results, given the nature of convention bounce.

Convention bounce is, in the simplest terms, the huge leap in polling that a candidate typically sees shortly following his or her party’s National Convention. The name “bounce” is chosen because, like a tennis ball spiked hard at the ground, shortly after the meteoric rise comes a steady fall.

As for why the bounce is a common occurrence year after year, a large factor is party unity. After such a high-energy and highly-viewed televised event, both parties typically enjoy a resurgence of interest in politics from viewers, resulting in higher polling numbers. By this point in the election season, post-convention polling numbers have typically regressed back towards the mean.

Convention bounces can mislead the public, as they’re not discussed in depth on the news or typically mentioned at all when looking at raw polling data. Candidates Walter Mondale and Al Gore both saw large bumps in their polling numbers shortly after their respective conventions. Subsequently, both went on to then lose the general election.

It’s easy to conclude, then, that the size of an election bounce doesn’t necessarily coincide with the favorite candidate to win the election, but perhaps more so how stable the polls are. According to 538, the volatility of the polls just preceding the convention is typically a good indicator of just how big a convention bounce a candidate may get.