Where Do Stocks Go From Here? History Provides a Playbook

Correlations are sky-high between 2020 and these two other periods.
Author:
Publish date:

Stocks are bouncing for a third straight session Thursday, something that’s become something of a rare event - the S&P 500 hasn’t seen three consecutive higher trading days since February.

Is the market finally putting in a bottom? Or are we seeing another bear market rally before the next leg lower?

Covid-19 may be without precedent in a lot of respects, but the good news is that history provides something resembling a playbook for investors right now.

To figure out the likeliest next step for the stock market, we’re turning to the data to see what we can glean from statistically similar timeframes.

Despite a surge in volatility, the price action in the big market indexes has been relatively technically obedient. Two weeks ago, the S&P 500 was in make-or-break mode, testing its long-term trendline right below the 2,800 level. Sure enough, that trendline broke, opening the door to another 19% downside move.

This selloff has been distinct in that it’s one of only three times in stock market history when the market has crashed from all-time highs. The other two were 1929 and 1987.

It’s more than a passing comparison however. Correlations between all three periods are extremely high:

1_DD_from_ATH_ps

To paraphrase the old Mark Twain quote - history doesn’t repeat but it often rhymes. And it’s rhyming now.

Why do market analogs matter? In periods of extreme volatility, statistically similar market environments provide a playbook for what might come next because of the way we’re wired. The latest findings in behavioral finance have shown that in times of psychological stress, humans react predictably.

So what do our other crash analogs say about what’s up ahead?

For starters, looking at all drawdowns that reached 30% or more - not just 1929 and 1987 - it’s clear that stocks generally find themselves higher in the months that follow, after some continued sideways chop:

1yr_forward_returns

Looking specifically at 1929 and 1987, the correlations are uncanny:

blended_return

To avoid affecting correlations, the chart above factors in the faster speed of 2020’s crash by speeding prior timeframes linearly by 1.7 times -  critically, the drawdown axis isn’t altered.

While the magnitude of the moves differs, the turning points line up incredibly well – both then and now. If those prior periods continue to hold up in the sessions ahead, we could be looking at an intermediate-term bottom.

Longer-term, there’s more of a question mark. While 1929 and 1987 had sky-high positive correlations in the first 100 days or so post-high, they decoupled within the year as 1929 got a second round of selling and 1987 avoided the double dip.

From a technical standpoint, it’s still too early to get aggressively long – the trendline break we warned about two weeks ago is still fresh.

But the data suggests it’s also wise to pare back aggressive shorts (and volatility longs) in this environment as we reach the stretch where crashes typically stabilize.

Then, expect 2020 to show us something completely different when it decouples from its analogs.

When that happens is an open question. Monitoring the rolling correlation between the two time-series is one of the best ways to know ahead of time  I’ll continue keeping an eye on it in these virtual pages.