by Charles Bilello, Pension Partners
“Markets don’t crash from all-time highs.”
This old saying is making the rounds again. Why?
The Dow just closed at an all-time high for the 8th consecutive trading day and 34th time in 2017.
That’s a lot of all-time highs, and it comes after the 122 all-time highs in 2013-2016. We haven’t seen this many all-time highs in consecutive years since the 1990’s.
Naturally, investors are feeling pretty good, with one prominent strategist proclaiming today that the bull market could “continue forever.”
Interesting, but what does the saying “markets don’t crash from all-time highs” actually mean? And why do people say it?
It’s supposed to imply that buying all-time highs is safer than buying on any other random day. That because tops are said to take time to build, so there will be plenty of time to “stop dancing” before the music stops. It also implies that the largest declines do not immediately follow all-time highs, because volatility tends to be lower on all-time high days than your average day. As volatility exhibits “clustering,” Benoit Mandelbrot observed that “large changes tend to be followed by large changes, and small changes tend to be followed by small changes.”
Are these assumptions supported by the data? Let’s find out…
First, we need to reaffirm that there is nothing inherently bearish about all-time highs. I’ve covered this many times in the past few years with the simple conclusion that “all-time highs tend to be followed by more all-time highs.”
That’s because, on average, forward returns are positive after all-time highs, meaning more all-time highs. The positive forward returns also happen to be a feature of any other random day, though marginally lower than all-time high days.
Note: all data herein are price returns only, not including dividends.
As for the percentage of positive returns, here too we find a likelihood of further gains, with a small edge to all-time highs in periods less than a year.
What about the risk of loss? An all-time high does not seem to preclude large future losses.
However, the largest losses following all-time highs pale in comparison to other days. Why is that? Because record highs tend to occur in periods of lower volatility. Since 1900, the Dow’s volatility on all-time high days was less than half of its average on all days.
It often takes some time for the market to transition from a period of lower volatility (where declines are small) to a period of higher volatility (where declines are large). Hence the saying: “markets don’t crash from all-time highs.”
But “some time” does not necessarily mean a long time. In fact, the worst crashes in the last 100 years, 1929 and 1987, both occurred within two months of an all-time high…
- On September 3rd, 1929, the Dow hit its 33rd all-time high of the year. Over the next month, it would fall 10% and over the next 3 months it would fall 36%. On October 28 and 29, it fell 12.8% and 11.7% in back-to-back sessions. By the time it bottomed in July 1932, it had suffered an 89% decline. There wouldn’t be another all-time high until 1954.
- On August 25, 1987, the Dow hit its 56th all-time high of the year. Over the next 3 months, it would fall 29%, with the largest one-day decline in the last 100 years occurring on October 19 (-22.6%). The Dow wouldn’t hit a new high again until August 1989.
Crashes get all of the attention, but most bear markets are not like 1929 or 1987.
- On February 9, 1966, the Dow hit an all-time high. It declined 25% before bottoming that October. A new high was not reached again until 1972.