5 ms·

This is an analysis of U.S. stock market returns over the past 150 years. A few insights: The average return of the U.S. stock market has been 8.4% per year o

This is an analysis of U.S. stock market returns over the past 150 years. A few insights: The average return of the U.S. stock market has been 8.4% per year over the past 151 years (1871 to 2022); this is the "real total return" reflecting dividends and inflation While the U.S. stock market has trended upwards over time, the market has declined in 31% of all years on record (47 years out of 151 years in total); for example: in 2022, the U.S. stock market dropped by 23.3% The range of returns across 1-year periods has varied significantly (from negative 37.0% to +53.2%). However, the annualized returns across 20-year periods have a much tighter range (from +0.5% to +13.2%) In other words, the stock market has never declined over any 20-year time period! Sources: Professor Robert Shiller and Yahoo Finance; note: the “U.S. stock market” refers to the S&P Composite index from 1871 to 1957, and the S&P 500 index from 1957 until today

retube 4y ago
> The range of returns across 1-year periods has varied significantly (from negative 37.0% to +53.2%). However, the annualized returns across 20-year periods have a much tighter range (from +0.5% to +13.2%) You'd expect something like this. For a normally distributed iid, the annualised volatility of returns over n years scales as sigma / root(n). So if your one year vol was 10%, the annualised vol over a 20 year period would be 10%/sqrt(20) = 2.23%.
- xapata 4y ago
  Are they normally distributed?
  - pclmulqdq 4y ago
    Almost, but the distribution has slightly fat tails.
    - ttymck 4y ago
      Is it correct to say that makes it (slightly) more like a uniform distribution, if viewed as a spectrum between one-point distribution and uniform?
      - xapata 4y ago
        It's hard to imagine a uniform distribution over an infinite domain, but that sounds right if you're think of it as a half-open spectrum.
  - kqr 4y ago
    I don't think normality is required -- the sqrt(n) scaling factor comes out of variance laws and the definition of the mean. It should be true for any distribution that has a variance, and the 150-year historic return certainly has a variance.
    - xapata 4y ago
      The sample variance is different from the distribution variance, but I get your point.
fedeb95 4y ago
"In other words, the stock market has never declined over any 20-year time period!" That's not true. It is if you cheat by taking useless averages. It as declined, as you say too, a lot of times, and very badly, many many times.
- pc86 4y ago
  There doesn't appear to be any period in question where you could invest in a broad-based index of funds and withdraw that investment 20 years later at a loss. That's what is shown in the FA and what the comment above means. Nobody's saying it has never declined. What isn't true about the above statement? It's incredibly specific, yes, but it shows that at least historically buying and holding over time limits huge gains but also limits losses.
  - fedeb95 4y ago
    It doesn't limit anything. Let's say you put 100k in the market today. With probability p you will have 8% return at some unknown point in time in the future (your argument may be false even from a simple temporal point of view, take a day in which the market plummeted, and take the closing price of that day, not the end of the year value, and 20 years prior I doubt you would have made a profit, but I can't make any statistics as of now). That means your 8% is unknown. So you haven't really mitigated any loss. The distribution of returns is unknown, at any time period, dayly, yearly, whatever, so you can't know your p, so any wishful calculation is nothing. Now, let's assume what I said up to now is wrong. I don't pretend to know the absolute truth anyway. Reflect on this: even if something has made a profit buying and holding for 20 years consistently for a millennia, it means nothing for my future. What guarantees me that I will make my profit in my 20 years? I can't travel back in time. Before discovering black swans, everyone thought swans could only be white. Also, 150 years of data on a phenomenon that changes almost daily is nothing.
- getToTheChopin 4y ago
  The linked page covers this in detail. The market goes down often, and sometimes goes down significantly for several years! The specific point being made is that there has never been a 20-year period where the U.S. stock market declined -- when comparing the start versus end value of the S&P 500 index, on a dividend / inflation adjusted basis.
albert_e 4y ago
if I invest for say 10 years with a conditional exit/hold strategy at tne end (i book profit and exit if I am in green, or decided to wait up to 5 more years till i turn green / with a minimum threshold) ...would that flexibility in investing strategy bump up my annualized returns (and presumably significantly reduce negative returns)?
- pc86 4y ago
  This doesn't sound like something you can know without back-testing that assumption (which is fairly easy).
- getToTheChopin 4y ago
  You can run a back-test on that type of strategy using the underlying data from the linked post: https://drive.google.com/drive/folders/1hacdyPFJtLMybJrf4CwFvmb0w8CgE3re https://drive.google.com/drive/folders/1hacdyPFJtLMybJrf4CwF... Edit: typo
- kqr 4y ago
  This sounds at its core like a "wait until I've made my money back" strategy. And yes, per definition you're guaranteed to make your money back if you follow it. It's not E log X-optimal though.

retube 4y ago

> The range of returns across 1-year periods has varied significantly (from negative 37.0% to +53.2%). However, the annualized returns across 20-year periods have a much tighter range (from +0.5% to +13.2%) You'd expect something like this. For a normally distributed iid, the annualised volatility of returns over n years scales as sigma / root(n). So if your one year vol was 10%, the annualised vol over a 20 year period would be 10%/sqrt(20) = 2.23%.

xapata 4y ago
Are they normally distributed?
- pclmulqdq 4y ago
  Almost, but the distribution has slightly fat tails.
  - ttymck 4y ago
    Is it correct to say that makes it (slightly) more like a uniform distribution, if viewed as a spectrum between one-point distribution and uniform?
    - xapata 4y ago
      It's hard to imagine a uniform distribution over an infinite domain, but that sounds right if you're think of it as a half-open spectrum.
- kqr 4y ago
  I don't think normality is required -- the sqrt(n) scaling factor comes out of variance laws and the definition of the mean. It should be true for any distribution that has a variance, and the 150-year historic return certainly has a variance.
  - xapata 4y ago
    The sample variance is different from the distribution variance, but I get your point.

fedeb95 4y ago

"In other words, the stock market has never declined over any 20-year time period!" That's not true. It is if you cheat by taking useless averages. It as declined, as you say too, a lot of times, and very badly, many many times.

pc86 4y ago
There doesn't appear to be any period in question where you could invest in a broad-based index of funds and withdraw that investment 20 years later at a loss. That's what is shown in the FA and what the comment above means. Nobody's saying it has never declined. What isn't true about the above statement? It's incredibly specific, yes, but it shows that at least historically buying and holding over time limits huge gains but also limits losses.
- fedeb95 4y ago
  It doesn't limit anything. Let's say you put 100k in the market today. With probability p you will have 8% return at some unknown point in time in the future (your argument may be false even from a simple temporal point of view, take a day in which the market plummeted, and take the closing price of that day, not the end of the year value, and 20 years prior I doubt you would have made a profit, but I can't make any statistics as of now). That means your 8% is unknown. So you haven't really mitigated any loss. The distribution of returns is unknown, at any time period, dayly, yearly, whatever, so you can't know your p, so any wishful calculation is nothing. Now, let's assume what I said up to now is wrong. I don't pretend to know the absolute truth anyway. Reflect on this: even if something has made a profit buying and holding for 20 years consistently for a millennia, it means nothing for my future. What guarantees me that I will make my profit in my 20 years? I can't travel back in time. Before discovering black swans, everyone thought swans could only be white. Also, 150 years of data on a phenomenon that changes almost daily is nothing.
getToTheChopin 4y ago
The linked page covers this in detail. The market goes down often, and sometimes goes down significantly for several years! The specific point being made is that there has never been a 20-year period where the U.S. stock market declined -- when comparing the start versus end value of the S&P 500 index, on a dividend / inflation adjusted basis.

albert_e 4y ago

if I invest for say 10 years with a conditional exit/hold strategy at tne end (i book profit and exit if I am in green, or decided to wait up to 5 more years till i turn green / with a minimum threshold) ...would that flexibility in investing strategy bump up my annualized returns (and presumably significantly reduce negative returns)?

pc86 4y ago
This doesn't sound like something you can know without back-testing that assumption (which is fairly easy).
getToTheChopin 4y ago
You can run a back-test on that type of strategy using the underlying data from the linked post: https://drive.google.com/drive/folders/1hacdyPFJtLMybJrf4CwFvmb0w8CgE3re https://drive.google.com/drive/folders/1hacdyPFJtLMybJrf4CwF... Edit: typo
kqr 4y ago
This sounds at its core like a "wait until I've made my money back" strategy. And yes, per definition you're guaranteed to make your money back if you follow it. It's not E log X-optimal though.