Historically the gold standard for predicting recession, more accurate than any other indicator, is "10-Year Treasury Constant Maturity Minus 2-Year Treasury Constant Maturity (T10Y2Y)":

https://fred.stlouisfed.org/series/T10Y2Y/

The gray bars are recessions and downward slopes precede recessions. At least all the way back 1976.

Has anything fundamentally changed that makes this no longer accurate?

EDIT: this comment has been downvoted with no explanation why. Perhaps the HN economist who downvoted me would be kind enough to teach us all the flaws of this predictor?