I've got some data-based takes about #election2020 I wanna share.

#1 The US polling industry is a mess. Despite some exceedingly good firms - and *there are many*, there's enough bad apples who douse themselves in false transparency and bullshit methodology to ruin it for the rest of 'em. UK polling cert not perfect (), but Christ, America.

#2 Hispanics are notoriously hard to poll. Seeking to entice a disengaged electorate where the language barrier often makes an appearance risks unrepresentative samples of likely voters. Owing to noticeable swings since 2016 (sometimes in Trump's favour), could error either way.

#3 In 2016 Trump improved support among voting Hispanics by 2pts. Polls show a more engaged Hispanic popilation this time round, and suggest Trump is making even further net gains. Certainly possible, but keep #2 in mind when reading comments which highlight this.

#4 With both #2 and #3 in mind, were I to make a bet it'd be that the rust belt will fall to Biden easier than the sun belt. White voters without a college education (particularly women) have been showing some of the strongest and most consistent swings to Biden this campaign.

#5 The RCP average is questionable. At best.