Stock Returns After Periods of Above Average Performance

Many investors have been operating under the assumption that we’re due for a market crash simply because stock returns have been so high over the past five years or so. The two large crashes we’ve experienced in this century alone have turned some forecasters into the financial prognosticator equivalent of Pavlov’s dogs.

Ring the bell (high returns) and the dogs salivate (call for a market crash).

Market crashes are always possible, but they’re rare. A higher probability forecast would be that we’ll see lower returns in the future (another prediction that’s impossible to time). In theory, this one makes much more sense than a crash since periods of relative above average performance are normally followed by periods of below average performance. But nothing is a sure thing in the markets. It’s always the timing that gets you on these types of market calls.

The S&P 500 is up around 16% per year over the previous 5 years. These may seem like extraordinary returns, but it’s really not out of the ordinary. Actual market returns are rarely close to the average. Here’s the performance breakdown of the entire U.S. stock market by rolling monthly periods going back to 1926:

10 yr rtnsThe average annual return over this period is 9.9%. This data shows that if you picked any month going back almost 90 years, there was around a one in five chance that 5 years later your annual returns would have been between 15-20%. A little more than one out of every ten times the market had negative annual returns over a 5 year period, about the same odds of finding returns over 20% per year.

The ten year numbers are still spread out, but are more clustered that the 5 year results as the extreme gains and losses are reduced:

5 yr rtns

Of course, it’s easy to look back at the last five years and show what has happened. Things always look easy with the benefit of hindsight. The question is: What happens after we’ve experienced huge gains in the market?

Here are the 5 years returns broken out the same way as they are above, except now I’ve calculated the subsequent 5 year returns:

5 yr sub rtns

What this tells you is that if the previous 5 years of performance was between 15-20%, the average of the following 5 years was over 13% annually. Not really what you would expect, right? As a reference point, the average 5 year return was 9.7% annually. But it’s worth pointing out that there is a wide range of returns from the best to worst. Over any 5 year period, the results can be all over the place.

The 10 year returns show more consistency in the range of outcomes:

10 yr sub rtns

You can see that the worst 10 years returns are a big improvement over the worst five years returns. Surprisingly, even periods of decade-long above average performance have been followed by 9-10% annual gains on average. The average 10 year return over this time was 10.4% per year.

Based on this historical data, investors aren’t doomed to experience poor market returns going forward just because we’ve had solid performance in the recent past. Obviously, everything is circumstantial with the markets so anything is possible. That’s part of what makes investing so interesting and frustrating all at the same time. You can’t predict the future with any precision by looking exclusively at past data. There’s a caveat for every rule and market data point.

While anything is always possible, there are some patterns in these numbers that investors can use to increase their probability for success, which is the best anyone can hope for. For any long-term investor in the stock market, there are two very basic ways to improve your returns:

(1) Buy low after there’s been a market crash.

(2) Increase your holding period.

Just look at the differences in the results for the 5 and 10 year data. Nothing is promised to investors, but the range of outcomes improves substantially as the holding period increases. The probability of loss goes down. Extreme outlier events decrease.

This doesn’t mean there are no losses during these longer time frames. Nothing works all the time. There will be periods in the stock market that are frustrating because the market won’t really go anywhere or you could even lose money over decade-long stretches. These things happen, but they are completely outside of your control.

What you do control is your time horizon and time is often the most important variable to understand when making investment decisions.

Subscribe to receive email updates and my monthly newsletter by clicking here.

Follow me on Twitter: @awealthofcs

Download PDF

Full Disclosure: Nothing on this site should ever be considered to be advice, research or an invitation to buy or sell any securities, please see my Terms & Conditions page for a full disclaimer.

What's been said:

Discussions found on the web
  1. John commented on Nov 12

    Awesome blog and one of the few I read regularly.

    Question: Would this theory be substantiated by similar data even for the Japanese market? Just trying to figure out how to avoid home country bias and avoid situations where the poor Japanese have not had reversion to mean in a long time.

  2. Mr. Frugalwoods commented on Nov 12

    It really is all about time horizon. And diversification.

    In response to the question about home country bias above, I think good diversification can help you avoid that. Make sure you have ex-US international exposure in your portfolio.

    Although, let’s be honest. If we had a period of doldrums like the Japanese have endured… we’d have bigger problems as a society than lagging retirement accounts.

  3. john kageleiry commented on Nov 12

    What would be an interesting overlay on this would be to incorporate the CAPE or some valuation metric at the end of the first 10 year return number, in this article, 16%. Where we are in the low-mid twenties in that CAPE, it projects far lower outcomes for stock returns over the next 10 years. Hopefully we would be able to more closely capture the probability of getting worst case returns vs average returns.

  4. Mark Zoril commented on Nov 12

    Very, very useful information!

    • Ben commented on Nov 12

      Thanks Mark.

  5. Alex commented on Nov 12

    Does not this contradict 10y market returns based on CAPE Shiller??

    • Ben commented on Nov 12

      See comment above. No easy answer.

  6. Marc commented on Nov 12

    This is a really nice article. The variability of subsequent returns shows concisely and with solid data why those technical analysts who use past market performance to predict future movements are doomed to fail most of the time!

    • Ben commented on Nov 12

      Really any past data is worthless as a way to predict future performance with any certainty (unfortunately). The best use of historical data is to define risk and loss possibilities. I think the people that use technicals in the right way use it as a form of risk management. The ones who think it helps predict the futrure are doomed to fail just like anyone else that tries to guess what’s going to happen.

  7. 10 Thursday AM Reads | The Big Picture commented on Nov 13

    […] Stock Returns After Periods of Above Average Performance (Wealth of Common Sense) see also 52-Week High and Momentum Investing (Alpha Architect) • Michael Lewis: Extreme Wealth […]

  8. V commented on Nov 13

    Sorry, the analysis is flawed. In your research you have created rolling windows of 5 year or ten year returns that shift every month. Thus returns in different rolling windows are similar or overlapping. Then you go ahead and calculate realized returns across this various rolling windows samples. The samples are not independent. It is like double counting. Your analysis can make statistically correct statements about each individual sample but when you create descriptive statistics across these samples you are double counting. Descriptive statistics are wrong.

    • Ben commented on Nov 13

      The analysis isn’t flawed because I never claimed they are independent periods. Nothing is wrong, only your interpretation of them. The sample size of independent data over a 90 year window is tiny. There’s no perfect way to analyze this data. And the independent data is nearly identical to this. Plus, normal people save an invest their money on a monthly basis over rolling periods. This is not a PhD dissertation, simply a different way to look at the data.

      • V commented on Nov 13

        Your post has a predictive tone to it. You are insinuating that if one were to hold a portfolio for 5 years what is the distribution of expected returns one can see.

        This is meaningful if returns are independent. Overlapping returns make for flawed samples. This is not the distribution one will face when holding a portfolio for 5 years. The whole thesis is wrong. Yes, independent returns lead to a small data set but that does not somehow condone the misrepresentation overlapped returns create.

        • Ben commented on Nov 13

          This is your interpretation of this. I’m simply presenting the data as to what’s happened in the past. You are drawing your own conclusions. I say anything is possible, but historically, here’s what has happened. You can make your own statistical conclusions based on the data.

          It doesn’t matter if the data are overlapping or not if someone is putting money to work in the markets on a regular basis. Each of those purchases would constitute an independent time horizon and I’m presenting the average of each of them.

          Statistics are clean, but the real world in the markets is messy. There are no easy answers. This is just one way to look at things.

          • V commented on Nov 13

            That I will give you. You should make very explicit what kind of returns to expect if one diligently invested every month. The results are the expected distribution faced by such an investor. But what your post says is that since we have been in a high returns market, what kind of returns to expect going forward while you let this fictitious and bold investor keep pumping his money as prices come down. This is not the same thing as holding a portfolio bought today for next 5 years. Please fix these things.

  9. Market Map commented on Nov 13

    We find that there is a constant flow of quantitative, single variable stock market related studies presented daily in the media, encompassing varying sample sizes, that can ultimately serve in undefinitive conclusions. The key to producing alpha is culling and “settling on” a multi-set of statistically significant factors from the vast stream of quantitative data presented, constructing a cohesive non-subjective model from them, and then letting it run for 15+ years. This is not easy.

    In a tactical realm, risk mitigation becomes more important the higher the market rises ( especially for retirees ). Using historical probabilities in terms of quantifying the market’s degree of “overperformance”, helps solidify the process of making of asset allocation decisions towards safety ( or buying assets during market “underperformance”) instead of acting on and relying on behavioral instincts or subjective interpretations.

    Also, we believe that a 90 year sample isn’t “tiny”. It has encompassed a large variety of economic and market environments, singular to this country’s election cycle, fiscal, and monetary idiosyncrasies. http://stockmarketmap.wordpress.com/

    • Marc commented on Nov 14

      Did you even read this article? The “Market Map” is a perfect example of what Ben is illustrating in this article as a completely flawed approach. It uses past performance and mean variance/regression calculations to “predict” the future. As Ben writes “You can’t predict the future with any precision by looking exclusively at past data.” The approach it uses ignores, for example, the fact (as Ben notes) that if the previous 5 years of performance was between 15-20%, the wide range of returns from the best to worst is from a positive 24% to a negative 8%. Yet the Market Map erroneously tries to use algorithms that analyze historical data periods of significant market underperformance to gain favorable entry price. To make note of Market Map after this particular article actually makes the Market Map look silly.

      • Ben commented on Nov 14

        Thanks Marc. My feeling is that even with 5,000 years of historical data we still wouldn’t be able to predict the future of the market because there will always be something different that comes about. There’s no such thing as perfect data, only how we choose to interpret it.

      • marketmap commented on Nov 18

        Tenets of disciplined portfolio management after many decades of experience :
        We will rigorously follow our model until the market (and not someone’s opinion) disproves it.
        We will diversify across proven, alpha generating stock universes (small cap value, emerging market small cap value, small cap momentum, Nasdaq 100 index, utilities portfolio.
        We will use low expense investment vehicles with which to generate alpha from those universes.
        We will not use leverage.
        We will not be swayed by subjective interpretation and behaviorally biased whims.

  10. ETF Cycles Review | Prudent Trader commented on Nov 14

    […] Bogle Math On Investment Returns (Servo) Stock Returns After Periods of Above Average Performance (Wealth of Common Sense) Gold Sentiment – A Contarian’s Dream? (Acting Man) What the stock market can teach you about […]

  11. Jacob Adom commented on Feb 17

    Knowing when to invest in the stock market is as important
    as knowing how much to invest. There’s never a guarantee that you’ll always win
    in the stock market game but with some new and effective tips and strategies,
    you can always earn loads of money. For more such visits, visit otcbully.com