The opposite approach is taken by a group of academics and consultants focused on politics: They prefer blunt, brute-force models that forecast elections using only a handful of factors.
Nate Silver, whose models for FiveThirtyEight rely heavily on polls, is the most famous of this group, but some of the forecasters work on a level even more removed from day-to-day detail and political personalities, looking at questions such as whether there's an incumbent in the race, or a recession in the offing.
The fact that some of these models perform pretty well has not discouraged the commentariat from pursuing its hobby.
And now, with President Trump, we who tell stories about the electorate may have an opportunity to beat the abstract indicators.
As it stands now, many narrow-factor models predict Trump winning in a landslide. He is an incumbent, and voters seem to generally prefer the devil they know to the one they don't. He is also presiding over a strong economy, and voters seem to be particularly reluctant to toss out the devil-in-residence during good economic times.
You can protest that presidents don't make the economy grow, and you'd be right. Presidents can play ham-fisted havoc with the economy, but they have a negligible ability to deliver upside surprises. Yet it's obvious that voters think they can, which means that any incumbent whose first term happens to coincide with a boom goes into a reelection campaign is heavily favored.
That axiom has only one caveat: Donald Trump. His approval ratings are far from where we'd expect them to be, given the strength of the economy.
Only one president has ever been in this ballpark at this point in his first term and gone on to reelection. And that president was Ronald Reagan, at a time when the United States was just beginning to pull out of a brutal recession.
Reagan had enjoyed an average 57 percent approval rating his first year in office, and after a rough couple of years post-recession, his approval ratings began rising back toward what we might think of as their natural point, peaking at 60 percent in 1986.
Trump, by contrast, entered office with a 45 percent approval rating, according to Gallup, a high point he has matched only once since then. It is currently bouncing between the high 30s and the low 40s.
It seems possible that the models predicting Trump's reelection, however well they've worked in the past, may contain a hidden assumption: a normal sort of president who will not repel swing voters with intemperate vulgarities or disappoint the base by not really pursuing a policy agenda. Which means narrative may beat predictive numbers in the next election.
But as we begin to construct those narratives, we should remember that journalists are apt to undervalue the strong economy because the economy for the news business isn't good.
Journalism's business model faces an existential threat, and the pervasive sense of economic gloom in the industry can color the reporting on what is in truth a generally healthy economy.
For the rest of the country, unemployment is low, and workers who gave up and exited the labor force between 2010 and 2014 are clearly being drawn back in by the lure of open jobs and briskly increasing hourly earnings.
Gross domestic product growth is strong, inflation is quiescent, and the stock market is near its all-time peak. Those things could change of course, but until they do, we shouldn't let our own troubles -- or our tendency to assume that every single thing Trump says is a lie -- blind us to the truth of economic growth under his administration.
Nor should we let our dour outlook tempt us to give those blunt, brute-force models shorter shrift than they deserve. It is highly possible that the 2020 campaign will see those simple models bested by grand narratives and complex stories.
But if we let our own circumstances dominate those narratives, we're likely to wind up surprised on Election Day, having succumbed -- as we did three years ago -- to blunt-force trauma.
Every weekday JewishWorldReview.com publishes what many in the media and Washington consider "must-reading". Sign up for the daily JWR update. It's free. Just click here.