Actual vs Estimated EPS

I was wondering if market multiples for 2025 might be understated because "companies sandbag" and EPS estimates just march higher over time - so that by the time we reach January of next year, the current 2025 SPY 19x EPS multiple ends up actually being 17-18x, etc.

Yeah - turns out that's not the case at all historically.

Over the past 10 years, the average difference between actual EPS in any given year vs the estimate at the beginning of that year is:

$SPDR S&P 500 ETF(SPY.US)$ : -2% (ie comes in below)
$Invesco QQQ Trust(QQQ.US)$ : 1%
$iShares Expanded Tech-Software Sector ETF(IGV.US)$ : -1%
$VanEck Semiconductor ETF(SMH.US)$ : 10%

The range is relatively tight in most years excluding the anomalous year (2020 / 2021). Interesting that it turns out rolled-up company estimates actually end up being about right in aggregate.

Also interesting that semiconductors have beaten estimates so handily. (Why?)

Very interesting that market returns in any given year are only somewhat correlated with the amount of EPS revisions in that year.

In fact, the best correlation I could find was the market has negatively correlated returns with the prior year's revision. So a larger likelihood of less interesting price returns in the year after upward EPS revisions, and a more interesting price returns in the year after downward EPS revisions.

Another reminder that macro is very tough and easier to just buy when people panicked / valuations low, and sell when everyone posting returns / valuations high. Also interesting that over time, bottoms-up estimates for EPS for the benchmarks winds up being about right.