Less than 1% of weeks that closed near all-time highs were followed by a 10% drop.

The S&P 500 closed the week at 7757.64, just 0.46% below its all-time high of 7793.68 with the VIX below 15. The usual weekend reaction followed: this must be the top.

This claim is testable. Since 1990, 242 weeks have closed within 1% of all-time highs with the VIX below 15. If markets near their highs were a real danger sign, the following month should have produced more large declines than a typical month.

A 10% decline within a month followed only 2 of the 242 weeks.

For each qualifying week, we measured the deepest drawdown the S&P 500 reached at any point over the next 21 sessions from that Friday close, and compared it with the same measurement taken from every week since 1990.

Only 2 of the 242 weeks that closed near an all-time high with the VIX below 15 were followed by a decline of 10% or more within the next month (0.8%). Across every week since 1990 the same threshold was reached 77 times out of 1,905 (4.0%). A week closing near the highs has therefore been about one-fifth as likely to precede a 10% drop as an ordinary week.

The two exceptions

There are only two cases to list: the weeks ending 26 January 2018 (−10.16%) and 14 February 2020 (−29.41%). So 36 years produced just two distinct events.

The risk is not zero, but it is extremely rare.

Why the probability of a big crash is only 0.8%

The reason is that when markets close near all-time highs, it reflects underlying momentum that historically tends to continue higher rather than suddenly reverse. Therefore, real crash risks near all-time highs are driven by black swan events or hidden vulnerabilities, not by the all-time high itself.