Sign up Log in
Sign up

Momentum crashes and the role of volatility

The Vola 6M column shows how strongly a price fluctuates. It is the only one of the four research columns that this list deliberately does not rate — no color, no sorting by it.

Momentum has a downside. Averaged over many years the effect is well documented, but the bad phases come rarely, clustered and hard. Research calls them momentum crashes. They do have to do with volatility — just not in the way one might first think.

This page explains when and why the crashes come, what scaling by volatility changes about them, why the column still carries no rating and how it is calculated. Where both sit in the history of momentum research is shown in The downside and the section on risk on the page on evolving the RSL strategy.

+232% loser decile in July and August 1932 — the market rose by 82% in the same months Kent Daniel 2016, text on Table 1
July/August 1932 the two worst months of the strategy, back to back, after a market decline of roughly 90% from the 1929 peak Kent Daniel 2016, text on Table 1
strong and volatile the stocks that did best in Levy's study — only in this combination Robert A. Levy 1967, Table 2

When momentum crashes: after the slump, in the rebound

The large losses do not come scattered but clustered — at the moment when the market rises strongly again after a slump.

Kent Daniel and Tobias Moskowitz analyzed the months from January 1927 to March 2013. The two worst months of the strategy are July and August 1932, back to back, after a market decline of roughly 90 percent from the 1929 peak. April and May 1933 rank 6th and 12th. March and April 2009 rank 7th and 4th; three of the ten worst months fall in 2009, in a three-month period in which the market rose dramatically and volatility fell.

The pattern is the same in both cases: first the market falls deeply, then it turns. The strategy is hit in the rebound. And the picture is not symmetric: the extreme losses are larger and more tightly clustered than the extreme gains.

The extreme momentum gains are not nearly as large in magnitude or as concentrated in time.

Kent Daniel 2016, text on Table 1

Source: Kent Daniel 2016, text on Table 1

Why: the losers rise

A momentum strategy does not crash because its winners fall, but because its losers rise.

The strategy used in research buys the winners and sells the losers short. When the market rises strongly after a deep fall, the stocks that rise most are precisely those that had lost most before — exactly the ones the strategy had bet would fall. The figures from Daniel and Moskowitz:

  • July and August 1932: Market +82%, Winner decile +32%, Loser decile +232%
  • March to May 2009: Market +26%, Loser decile +163%

Thus, to the extent that the strong momentum reversals we observe in the data can be characterized as a crash, they are a crash in which the short side of the portfolio—the losers—crash up, not down

Kent Daniel 2016, text on Table 1

For a ranking from which one only buys, this means: anyone who bought the top in 1932 did not crash. They were 50 percentage points behind the market — a poor result, but not a crash. The strategy's crash is the crash of its short positions.

It also means: the crash depends on the state of the market and on the volatility of the strategy as a whole, not on the volatility of an individual stock. How restless a particular stock is does not tell you whether a momentum crash is coming.

Source: Kent Daniel 2016, text on Table 1

The scaling: how much, not which

Barroso and Santa-Clara do not improve the selection of stocks. They change how much is put to work.

Pedro Barroso and Pedro Santa-Clara scale the whole strategy by the inverse of its own realized volatility over the last six months. When the strategy becomes restless, they put less to work; when it is calm, more. The crashes thereby practically disappear.

Three things about this matter, and all three argue against reading the column as “calm is good”:

  • What is measured is the volatility of the whole portfolio, not that of an individual stock.
  • The scaling works over time: the same strategy, sometimes with more, sometimes with less money at work. Which stocks it holds does not change.
  • The improvement comes not from a better forecast but from managing the amount at work. A ranking cannot scale anything.

Source: Pedro Barroso 2015, Abstract

Why the column is shown but not rated

When it comes to choosing stocks, Levy's finding speaks for volatility rather than against it — and even then only together with strength.

Robert Levy asked the question a ranking is about: which stocks? In his Table 2, the stocks that did best were those that were relatively strong and relatively volatile. Only in this combination; volatility alone predicts nothing there.

Levy 1967, Table 2Barroso/Santa-Clara 2015
Questionwhich stocks to choosehow much to put to work
Measured onindividual stocks in the cross-sectionthe whole strategy, over time
Findingstrong and volatile did bestcalm strategy: put more to work

The column measures the volatility of individual stocks. That is Levy's question, not that of Barroso and Santa-Clara. This is why in this list:

  • The column is not colored, neither green nor red.
  • It cannot be used for sorting, in either direction. A ranking by volatility alone has no evidence behind it — not “highest first” either.
  • The column's rank counts from the highest value, like all the others. That is a magnitude, not a judgement.

Levy's finding is a finding, not a rule. His sample was narrow — 200 stocks on the New York Stock Exchange from 1960 to 1965, without adjusting for risk — and that is exactly where Jensen and Benington's criticism took hold in 1970. Whether more recent work supports it has not been checked here and is therefore not on this page.

What the column does say for certain: a stock that fluctuates strongly can also fall strongly. High volatility means high risk, for a single stock as for a portfolio. What that means for investment decisions, and what this page expressly is not, is set out in the disclaimer.

Source: Robert A. Levy 1967, Table 2, Michael C. Jensen 1970, Summary and Conclusions, Pedro Barroso 2015, Abstract

How the column is calculated here

The standard deviation of daily returns over the last 126 trading days, annualized, in percent.

6 months at the usual approximation of 21 trading days per month make 126 trading days. 127 daily closing prices give 126 daily returns; their standard deviation is the volatility of one day. The calculation uses the sample standard deviation, i.e. n − 1 in the denominator. If a stock does not yet have 127 daily closing prices, the field stays empty.

It is annualized with the square root of 252, the usual number of trading days in a year. The square root rather than the multiple, because random fluctuations partly cancel out over the days: volatility grows with the square root of time, not with time itself.

A worked example you can check

A made-up price rises by 1.00% one day, falls by the same rate the next, and so on, for 126 trading days. The volatility of one day is then almost exactly that rate. Annualized:

1.00% × square root of 252 = 15.87%. The column shows 15.94% for this price; the small difference comes from n − 1 in the denominator.

Roughly, the value reads like this: a stock with a volatility of 15.87% moves by about 1.00% on an ordinary day — sometimes up, sometimes down. The number says nothing about the direction.

Not investment advice and not a recommendation. Past performance is not a reliable indicator of future results, and investing in securities involves the risk of losing your entire investment.

Disclaimer and risk information in detail

References

Every figure on this page is shown with its exact location in the source — section, table or page. The links lead to the paper via its DOI; the full texts are mostly behind publishers' paywalls, but the bibliographic details are enough for any library.