Leading indicators work because they capture decisions made now that produce activity later: a permit pulled today is construction in six months, an order booked today is production next quarter.
They come with two costs. Lead times vary enormously, so a signal can be early by a quarter or by two years, and false positives are common. The standard defence is to require several independent leading series to agree before acting, which is what composite indices attempt.
Example: the three-month-ten-year curve inverts, the four-week average of initial-jobless-claims rises 15%, and housing-starts permits fall 11%. Three independent leading signals agreeing is far stronger evidence than any one of them alone.
Related: lagging-indicator, initial-jobless-claims, housing-starts, three-month-ten-year, inverted-yield-curve