To see how CAPE effects the performance of buying during new market highs, you'll need to have a lot more samples. By only looking at data where the CAPE is above 35 you only have two significant sample periods (the first in 1998-2001) and the second in 2020-2021). Though you are testing back to 1950, the CAPE ratio being below your target nearly constantly is causing your sample size to be so small as to lose its utility.

I didn't use 35. I used existing 97th percentile based on previous values. So going forward, the CAPE value for the 97th percentile changed. For instance, in April 1964, the 97th value was 22.42.

However as Jim has pointed out, and I want to underline - the CAPE ratio isn't useful as a timing device, as it only works over 5-10 year periods, but it is excellent to narrow the range of expectations for you returns to help with your planning, not to mention your sanity.

Yes, I agree that CAPE is not useful for timing, but it seemed like some people in this thread were saying the CAPE is high, or other factors were high, look for other investments.

I looked at the forward 25 year values from CAPE quintiles. I thought 25 years would be a reasonable number for retirement planning.

The bottom of the returns (I thought best to look at the bottom returns for retirement planning) was 8.8% for the bottom two CAPE quintiles and 7.4% for the top two quintiles. Mid quintile was 8%. So for 25 year retirement planning, assume 7% with a 50% drop somewhere along the journey no matter what the CAPE value is.

The top of the 25 year range was 17% for the bottom quintile and 13% for the top quintile.

Aussi