Former President Donald Trump and President Joe Biden.
CNN  — 

If you follow my work, you know that I’ve tended to be pessimistic about President Joe Biden’s chances of reelection. There are plenty of reasons for that – for instance, he’s trailing in the polls both nationally and in swing states, and has an approval rating south of 40%.

But sometimes it’s worth taking a step back and looking at the counterargument. Biden’s been through arguably two of the worst weeks for a president running for reelection that I can recall, and he’s still within earshot of former President Donald Trump.

If you average the national polls since the debate 15 days ago, Trump’s ahead by 3 points.?An NPR/PBS News/Marist College poll out Friday morning even put the race at 50% for Biden to 48% for Trump (a result within the margin of error).

No one should be comfortable calling the presidential race with these sorts of polling results. Since 1972, the average difference between the polls at this point and the eventual result has been 6 points.

Sometimes races change by far more than that. Democrat Michael Dukakis led Republican George H.W. Bush by mid-to-high single digits at this point in 1988 before the party conventions. By the time the race was over, Bush had defeated Dukakis by 8 points.

Biden and Trump’s own history should make you think twice about calling wraps on this race as well. Biden was ahead by 9 points in the early July national polls in 2020. He ended up winning the national popular vote by only half that (4.5 points).

Such a shift in Biden’s direction this time around would put him over the top in the popular vote.

Of course, this election will ultimately come down to the pivotal battleground states. The post-debate data from those states is limited, though the polling we had from before the debate suggested that Biden was doing worse in those states taken as a whole than he was nationally.

Biden’s clearest path, according to the public data (and both campaigns), continues to be through the northern battleground states of Michigan, Pennsylvania and Wisconsin. Wins in all three would likely mean Biden can lose Arizona, George, Nevada and North Carolina and still pull off a victory.

Looking at the forecasting models

Figuring out how to account for all this information (or lack thereof) to better understand the state of the race between Biden and Trump can be a challenge.

One way to do that is through forecasting models, which look at a lot of data, including current national polls, current swing-state polls, the time until the election (i.e., how much polling can shift over time) and even the state of the economy (e.g., strong job reports and less positive data about income growth).

I love to check these forecasting models and have even assisted with building some in the past because they help ground us in objective facts and not subjective opinions. Models like these cannot account for everything (e.g., we’ve never had a rematch between two presidents in the polling era), but they are far better than conjecture.

Per the average forecasting model, Biden wins about 30% of the time against Trump. Some might view a 3-in-10 probability as low. I wouldn’t read it that way.

If the current model average is to be believed, Biden has a better chance of winning this election than a flipped coin?landing on heads twice in a row. Any young student or NFL captain can tell you that flipping heads twice in a row happens all the time, despite it not being the more likely outcome.

For the politically obsessed, a different example might hit closer to home. When I worked at FiveThirtyEight – my then-boss Nate Silver now has a model that looks similar to the current average – we gave Trump about a 30% chance of defeating Hillary Clinton on the eve of the 2016 election.

As I wrote at the time, Trump was a “normal polling error behind Clinton” – the idea being that the polls were close enough that Trump could win if the polling was off by a normal amount.

Today, the situation is somewhat different. Biden’s Electoral College path does seem to be narrower than Trump’s was in 2016.

What Biden does have that Trump didn’t in 2016 is time. We’ve already seen over the past few weeks since the debate that a lot can change in a short period.

So what do all these facts and figures mean? I think it’s totally fair to say that Trump is the favorite and that Biden has his work cut out for him. There is also data to suggest that Vice President Kamala Harris would be a stronger?Democratic nominee against Trump than Biden. (I tend to believe Democrats would do better with Harris leading the presidential ticket.)

But if you currently think Biden can’t win, history would disagree with you.