We live in an era of horribly misinformed political decision-making. In 2015, the Trump-skeptical Republican elite believed that Trump would naturally fade over the course of the primary, so they remained split between mostly implausible presidential candidates (looking at you, Jeb), and Trump won. That same year, major Democrats cleared the field for Hillary Clinton, who turned out to be the second most disliked presidential candidate ever and one of the few Democrats capable of losing an open race to Donald Trump.

Most of these mistakes were based on bad political calculations. People overestimated Clinton’s strength, the demographic resiliency of the Democratic coalition ..., and their plans fell apart. And both Republicans and Democrats are in danger of making new mistakes by underestimating Trump’s strength in 2020.

The conventional wisdom on Trump’s reelection odds are much too bearish. Trump has a better chance of winning again than many pundits seem to think. And both Democrats and Republicans need to take Trump seriously and recalibrate their decision process or risk facing serious consequences.

If Trump is in [the mid-to-high 40 percent satisfaction range], he has a solid chance — he was able to win over some voters who disliked him in 2016, and a Democratic nominee who runs on a far-left platform might drive some softer, Trump-disapproving Republicans back into the fold.

Approval ratings change — historically speaking, a president’s approval rating roughly 770 days from inauguration (which is where we were about a week ago) doesn’t have a strong relationship with their eventual reelection vote percentage.

To me, this situation adds up to Trump having roughly 50-50 odds of winning reelection. There are a lot of ways that things could go very wrong for Trump, but there are also a lot of ways things could improve enough to put him over the top. Other people might come with somewhat different math, but, regardless, it’s a risk to assume that Trump is a one-termer walking.

More From The Washington Post: