“Whenever you find yourself on the side of the majority, it is time to reform (or pause and reflect).” ― Mark Twain
This week is Blogger Wisdom week on Abnormal Returns. As we have done in previous years we asked an esteemed group of independent finance bloggers a series of (hopefully) provocative questions. Yesterday we asked what these bloggers know about investing (with confidence) and absent statistical support. The answers are not edited and the author’s name, blog title and Twitter handles follow. We hope you enjoy these posts as much as we do putting them together.
Question: If we had a 1,000 years of market data what kinds of things would get validated? What things would lack support in the data? (Answers in no particular order.)
I’m sure there are a ton of statistical relationships that we currently hold dear that would look foolish with this much data but I’m not exactly sure what that would be. I believe that markets are constantly changing and adapting based on our knowledge of the past and the technology that’s available to sift through all of the relevant market data. But I also feel that market cycles are shaped by our memories and personal experiences with the markets. It seems like newbies in the markets will always have to learn the key lessons on their own while experienced investors can have their views clouded by their own formative years. I think the only thing that would really get validated with 1000 years of market data would be that everything is cyclical and those cycles are driven by human emotions.
Tobias Carlisle, The Acquirer’s Multiple, @greenbackd, co-author of Concentrated Investing: Strategies of the World’s Greatest Concentrated Investors:
We get the same answers we have now but still no resolution. If we had 1,000 years of data, people would say, “Well, sure, but that data set started in the Dark Ages after the fall of Rome, through the rise of Western Civilisation and culminated with the Industrial Revolution. That’s why momentum stocks looked so strong. If we had a 10,000-year backtest, value might just start outperforming.” 10,000 years later value would still be waiting for its time. Any day now.
I think with 1,000 years of data we would be able to determine if valuation based market timing (either broad market or factors) is worth the effort. I think with 1,000 years of data we would find that many stock selection factors don’t work.
Buggy whip manufacturers would have been the greatest investment of all time–except they listened to consultants who told them that dividends were archaic and tax-inefficient, so they kept plowing earnings back into the business. When the companies went belly up and their shares became worthless, the multi-century return to investors was -100%.
[E-mail] Subscribe to our daily e-mail newsletter to stay up-to-date with all of our posts.
Even if the trend towards passive investing slows down, it’s hard to imagine it completely switching direction anytime soon. Many active managers have argued that too much cap-weighted indexing would be bad for markets, leading to massive distortions in the absence of their price-setting activities. Others, including AQR’s Cliff Asness and pseudonymous-blogger Jesse Livermore, have written at-length on the counter-argument – that an increase in indexing could make markets more efficient.
While we all (I think) would agree that zero active management would be a troubling (and odd) scenario, it remains to be seen whether or not there is such a thing as “too much passive.” A thousand years of data would likely see a real-world test as to whether or not there is an upper-bound to the amount of indexing that can occur while still allowing capital markets to properly function.
There are no “new eras” though investors inevitably fall in the love with the idea at every major peak.
All of the messy human parts of markets would be validated. Performance chasing and euphoria about the next new thing. Bubbles and despair would be among the most regularly repeated features of history. Manias, panics and crashes are a part of the human financial experience across time and location. Sharpe’s arithmetic of returns, fees and investor performance can’t be changed, and most people will have found themselves unable to capture the lion’s share of returns over those 1000 years.
The idea that stocks have a perfect value that they “should” revert to would be lost. The “correct” P/E ratio, CAPE, equity risk premium and what investors expect for future returns will never stabilize.
1000 years of data would validate the fact that we would still want 2000 years of data! I think it would invalidate things that happen on a lifetime horizon, such as US relative outperformance. Passive investing fanatics would love to be able to point out that a holding period of “all of human progress” is quite profitable.
Based upon my response to the first question (“performance distorts decision making”), I would expect that there would be value and momentum premia, but smaller than we believe to be available today.
[Membership] An annual membership helps support Abnormal Returns and gives you access to exclusive e-mails and perks.
If we had 1000 years of market data, we would probably be more confused than we are now. We often forget how much data series change over time. Industries go in and out of existence. Survivorship bias is difficult to deal with. Contracts used in investing change over time, as does their degree of liquidity and the degree of disclosure, regulation, etc. Cultures change over time as well; wars and economic crises change willingness to take risk, as does technological change.
What doesn’t change is the fundamental nature of man. Boom and bust, fear, greed and envy are our birthright. “Yet man is born to trouble, as the sparks fly upward.” [Job 5:7 NKJV]
Because of that, most of the basics of behavioral finance will be validated most of the time, leaving just enough doubt for people to wonder if the principles are really valid. Valuation, momentum, small size, and moderate credit risk will get rewarded on average. Other things will be more tenuous, and the ability to analyze the data will be difficult. That said, new versions of Triumph of the Optimists will likely get produced, because in the long run, moderate risk-taking wins.
No matter the country, no matter the decade, publicly traded stocks and securities have always been subject to the same
price swings, uptrends, downtrends, and buying or selling patterns. That is because human nature changes very slowly (if at all) and we are all still driven by the same senses of fear and greed that ruled our forefathers.
As Jesse Livermore said in Reminiscences of a Stock Operator, Another lesson I learned early is that there is nothing new in Wall Street. There can’t be because speculation is as old as the hills. Whatever happens in the stock market today has happened before and will happen again.”
You would think the effectiveness of most investing and trading strategies would be confirmed or disconfirmed with 1,000 years of data. At the same time, markets are adaptive and anything we learn from 1,000 years of data might be altered as investors incorporate that information into their decision rules.
Regardless of how much data is available, people will always reach different conclusions about the same set of data depending on their preference for minimizing Type I or Type II error. For me, I’m generally more concerned with implementing a bad idea than missing out a good one.
The equity premium, the value of diversification and the critical importance of a long-term perspective would all be validated. Pretty much everything else would not.
This is a great question! Nothing would be validated. Those skilled in research design frame hypotheses and identify relevant data. Most who are not skilled (and some who are) begin by getting as much data as possible and using all of it. We have a lot of poor research results right now because of a brilliant historical job of finding or reconstructing data from 1871. To summarize, even 100 years is too long to expect unchanging relationships.
A) Government spending doesn’t push markets higher. There is no such thing as pump priming. Wars don’t help markets or get us out of recessions.
B) People having control over a process is better than a decentralized market allocating resources based on preferences and cost/opportunity cost. No amount of data will convince people that markets aren’t like puppets on a string. They think an individual or bureaucrat is smarter than a market.
Brett Steenbarger, TraderFeed, @steenbab, author of Trading Psychology 2.0: From Best Practices to Best Processes:
I question the presence of universal patterns in market data. Participants change over time, and economic/political/monetary conditions change. It is not clear that strategies that work today would be illuminated by data hundreds or thousands of years old. For example, in the recent zero interest rate global environment, can we expect stocks and bonds to move as they have at prior periods when no such monetary policies were in place? I suspect what might be validated over the course of many years are human responses to market movements (i.e., the results of cognitive biases, fear, greed, etc) and the positive expected returns of acting counter to those human responses.
[2017 Campaign] You can help those in need of clean drinking water by taking part in our 2017 campaign.
Investors have always been irrational and emotions have been the driving force behind their actions taken.
Great question here. One of the biggest problems with virtually all financial time series is that the data gets questionable prior to around 1920s. Before the 1920s, there were very few stocks traded and only a tiny fraction of the population had any exposure to the stock market. So, we have less than 100 years of what I would call usable or representative data. With 1,000 years of data, we’d have a much better idea of what a “correct” market valuation would look like and what a “correct” equity risk premium should be.
It’s easy to argue that both the Great Depression and the Great Recession could have been worse than they were. But for our data sets they’re as bad as it gets. So if we had 1,000 years of apples-to-apples data I think we’d find periods of 30-50 years, or more, of misery. Which is important, because once you can argue that markets can sink for longer than anyone can wait, someone can rationally argue that no one should ever touch them.
1,000 years worth of market data would certify that the financial markets did something in the past and that no one knows precisely what that means for the future.
Wesley R. Gray, Ph.D., Alpha Architect, @alphaarchitect, co-author of Quantitative Momentum: A Practitioner’s Guide to Building a Momentum-Based Stock Selection System:
Investors chase short-term performance and make bad decisions. The equity risk premium is not 6%. More like 2-3%.
Momentum as a trading and investing system will survive the test of time.
In many ways, we do have 1000 years of data already when we are willing to look at capital markets outside the United States. Even with 1000 more years of data, I do not think much would change, the same way that 1000 more years of weather readings would not help us better predict whether it is going to snow in Boston next February 10th. There are just too many variables and too many non-linear relationships.
Insights would likely be on the big, long-run stuff. I think we’d find much the same. On the one hand, we’d continue to see that past performance is not indicative of future results. Returns vary. Markets do unpredictable things. Supposedly rare events happen with surprising frequency.
On the other hand, since past performance is really all the data we have to work with, more data could help us sharpen our statistical certainty around already well documented anomalies. I generally think we would continue to find that it is the simple stuff that is robust: buy cheap stuff and follow the trend.
Only the data could tell us that! But if I’m guessing, I’d guess the “behavior gap” and the value premium.
Over extended periods of time (perhaps even hundreds of years), certain factors often appear to be statistical valid, but few really were over the entire time frame. Buying stocks below their intrinsic value would be one factor that was valid and led to outperformance.
Thanks to everyone for their time and effort. The next edition of Blogger Wisdom touches on the question of what will embarrass us ten years in the future.