Skip to content

Defining and Dating Bull and Bear Markets: Two Centuries of Evidence: summary and key takeaways

The paper defines bull and bear markets as persistent shifts in the mean return and dates them in a long US stock index with two detection methods. Both methods find distinct return states that a constant-drift random walk does not explain. Simple timing rules built on the dates cut risk in the out-of-sample tests.

Gonzalez, L.; Hoang, P.; Massey, J.G.P. et al. · 2006 · Multinational Finance JournalSummary updated 20 November 2026
Published2006Journal article
Asset focusEquitiesBull/bear regimes
TypeResearchRegimes & Safe Havens
Reading time4 minthis summary

Defining and Dating Bull and Bear Markets: Two Centuries of Evidence at a glance

The question
Can bull and bear markets be defined and dated by formal rules as persistent mean-return regimes, and do the resulting states carry useful information for investors?
Dataset & period
Spliced US stock price index, monthly, January 1800 through December 2001. Schwert's series is used up to 1925, then S&P indices. Turning points use nominal capital returns. Trading rule tests cover January 1934 to December 2001 because of interest rate data limits.
Method
Bull and bear markets are modelled as alternating states in which the mean monthly return shifts, with each state required to be sustained. The BB method adapts the Bry-Boschan business-cycle algorithm: it finds local peaks and troughs and keeps phases with an absolute cumulative capital return of 10%. The CC method is a count-cusum procedure that compares returns in overlapping windows to a rolling median and flags switches between high-return and low-return states. Trading rules act on the detected turning points with a six month lag, moving between the index and Treasury bills, with a leveraged variant.

Key findings of Defining and Dating Bull and Bear Markets: Two Centuries of Evidence

  • Both methods place the start of famous bear markets, including 1929 and 1987, at the same points, and both capture the long 1990s advance.
  • Bull phases last close to 22 months on average under both methods. Bear phases average 15 (18) months under BB (CC).
  • The index sits in a bull state for just under 60% of months. BB and CC label a month the same way 84% of the time.
  • In bull states over 70% of months show a positive return, versus less than 40% in bear states. Volatility is only a little higher in bear states.
  • Where both methods agree on bull, mean monthly capital return is 2.10%. Where both agree on bear, it is –2.1%.
  • Out of sample, the conservative rule gives up 0.1% or 0.2% per month against buy and hold, with return volatility cut by close to a half. The leveraged rule earns 50% more than passive at comparable risk.

Robustness, as reported in the paper

  • Simulated random walks with constant drift (1000 simulations) produce phase durations that are too short compared with the historical BB durations. For CC the historical durations sit inside the simulated intervals.
  • Results from randomization of observations and of residuals match the parametric tests, according to the paper (not reported in detail).
  • Total returns give virtually identical results to capital returns, and the results do not depend on the phase length and cycle length rules (not reported).
  • Trading rules are tested over two sub-periods of equal length, and the paper notes that few trades occur so costs are unlikely to remove the benefit.

Caveats and limits

  • Turning points are only recognised after a six month lag, so the dates are known late. The conservative and leveraged rules therefore trade on delayed signals.
  • Nominal returns and an index spliced from different sources are used, and early data quality is limited. Dividends are unreliable for the early period.
  • Trading rule results come from one market over one 68-year window and no formal significance test of the rule returns is reported (inferred).
  • Only the US market is analysed in the paper. A British replication is mentioned but not reported.

Items marked (inferred) are our reading of the design, not statements by the authors.

Why it matters

The paper gives an explicit, rule-based way to label bull and bear states instead of relying on media dates. It shows that regime labels carry large differences in mean returns and that persistence is hard to reconcile with a constant-drift random walk. For risk work, it supports treating expected returns as state dependent.

Questions about this paper

How long do bull and bear markets last in this study?

Both methods put the average bull market near 22 months. The average bear market is 15 months under BB and 18 under CC.

How much of the time is the market in a bull state?

Under both methods the market is in a bull state for just under 60% of months. The two methods label 84% of months identically.

Do the bull and bear labels actually differ in returns?

Yes. Months both methods call bull average 2.10% and months both call bear average –2.1% in capital return.

Where can I read the original paper?

The original is at its publisher: https://doi.org/10.17578/10-1/2-3. We link to the paper and do not host the PDF.

Follow new quant research

Download MacroFinance market data as CSV or Excel, or pull it by API, and test the ideas in these papers yourself. New summaries are added as we review them.

Create a free account
Summaries are written by MacroFinance from the published papers, in our own words, for information only. They are not investment advice and may omit detail. Papers are not hosted here: each page links to the original at its publisher.