Detrended real total return factor is more predictive of future 10-year real returns than CAPE

Cyclically adjusted price-to-earnings (CAPE) is famous for its predictive power in estimating long-term returns. Recent multiple expansions have produced CAPE values more extreme than any time other than the dot-com bubble in the late 1990s. Here we show that detrending the real total return factor relative to a long-term exponential growth model is more predictive of future real returns than CAPE. This approach is oblivious to both earnings and the risk-free rate. We also show that the predictive power of CAPE is much decreased after removing this signal. Detrended real total return factor estimates valuations to be stretched relative to historical data but slightly less so than suggested by CAPE; even so, both CAPE and the method compared here estimate annualized future 10-year real returns beginning Feb 1, 2024 will be ∈ [2.656%, 2.973%] without conditioning on additional macro factors.