Do I follow political polls? Yes, but only after I’ve deconstructed the polling methodology and recalibrated the polls based on my best estimate of what they really say.

Let me explain. Polls are valuable in theory but flawed in practice. The flaws include oversampling voters of one party relative to their actual percentage in the electorate, oversampling minority groups with a disproportionate tendency to vote for one party over another, sampling “all voters” or “registered voters” instead of “likely voters” and other techniques that skew results to one party or another.

For example, Democrats have an edge over Republicans in registered voters. That edge (ignoring independents for this purpose) is about 54% to 46%. But pollsters survey a sample of 58% Democrats to 42% Republicans.

Within the Democratic oversample, there’s another oversample of minorities with a 90% track record of voting Democratic. That skew undersamples white Democrats who might vote Republican. There are many other examples, but you get the point.

The good news is that when you study the methodology and reverse-engineer the poll, you can come up with useful results for purposes of forecasting.

This article reports on a poll that exhibits some of these flaws (it polls “adults” rather than registered voters), but it still has its uses when adjustments are made.

Trump does much better in this poll than many others. What’s important analytically is not the spread but the trend. Voters are moving in Trump’s direction despite impeachment efforts.

Even a flawed poll has its uses if you put it in a time series of similar polls. Trump needs all the good news he can get these days, and this poll is one ray of hope.

Institutional investors can schedule a proof of concept with the world’s first predictive data analytics firm combining human and artificial intelligence with complexity science. Check out Jim Rickard’s company at Meraglim Holdings to learn more.