Large price runups, such as a gain of 100% over two years, are rare.1 In a previous blog post, I presented 10 attributes to distinguish asset class bubbles from large price runups that are justified by improving fundamentals. Those bubble attributes are:

  1. Heavy retail investor involvement
  2. New-era thinking
  3. Irrational valuations
  4. Five or more years of swiftly rising prices
  5. Parabolic rise in price
  6. Shorting is unattractive or impossible
  7. Social mania
  8. Product providers exploit excessive demand
  9. Leverage fuels more buying
  10. Bubbles are late-cycle phenomena

As asset class traders, we are especially interested in bubbles as a potential huge source of alpha when they collapse. As it turns out, bubbles are a lot tougher to exploit than it might seem. In this blog post, we’ll delve into bubble characteristics in more detail, and then investigate the best ways to trade asset classes that are experiencing a bubble.

We’ll examine bubble characteristics over the short term (plus and minus three years around the peak) and then longer term (a decade or more). When doing this sort of analysis, we need to at least acknowledge that various forms of hindsight bias can creep into such work since we are examining known price runups that ultimately crashed spectacularly.

It’s possible that we should include a few historical parabolic runups that did not ultimately pop.1 I’m hopeful that the lack of bubble attributes associated with these moves provides the justification for eliminating these from consideration, but I’m not completely sure. I may have also declared a few large up and down moves as bubbles (for instance, Chinese equities in 2007), when perhaps this price move had no more bubble attributes than a big move that ultimately didn’t end in a long period of underperformance (for example, the 1987 crash).

I’d like to acknowledge Alex Golubev, a fellow portfolio manager at Merriman, who was heavily involved with the bubble research we performed this year. This work is currently very topical because it appears that the cryptocurrency bubble is in the process of popping, having peaked in January 2018. Also, after almost a decade of excellent returns, perhaps U.S. technology stocks may eventually form into a bubble in the coming years.

Bubble Characteristics Around the Peak

First, we’ll examine bubbles that have daily price data to analyze. For some bubbles, especially those that occurred many decades ago, we use index data and return series that have been constructed many years later. Examples include using the Dow Jones Industrial Index and daily prices series from Ken French’s data library. For more recent bubbles, we can use actively traded exchange trade funds.

Table 1 shows the 11 bubbles we examined with daily data. We’ll be examining median and average properties. There are a variety of conflicting issues when selecting which bubbles to include in an aggregate study. Most were full-fledged bubbles listed in Table 1 of my previous post. Others, such as the oil price “bubble” of 2008 and the 1980s runups, had fewer bubble attributes and were perhaps a manifestation of excessive inflation fears, rather than extreme optimism associated with high future investment returns.

We can also debate possible triple counting 1980, with gold, silver and energy stocks all peaking in the same year and all responding to similar concerns of high inflation expectations. There may also be double counting associated with the year-2000 bubbles, and the Chinese Equities/Oil Price bubbles in 2007 and 2008, as roughly being associated with the same theme.

Interestingly, we’ve also added another bubble – bitcoin in 2013. Bitcoin was up around 85x from the beginning of the year to its peak on December 4, 2013. Consulting the list above, bitcoin’s 2013 bubble attributes are 1, 2, 4, 5, and 6. There’s never been a case of two bubbles in the same asset so close in time – which seems to have occurred with bitcoin in 2013 and 2018.

To account for our bubble selection difficulties, we’ll also look at stats associated with the “big 8” bubbles (eliminating oil, silver and bitcoin). These three are the smallest bubbles compared to a relevant GDP measure. Bitcoin is extremely volatile and was very small at the time, peaking at $14 billion in market value in December 2013. The silver spike was caused in part by the Hunt brothers’ manipulation activities and fears associated with high inflation. Eliminating silver also helps with the double/triple counting issue. Ultimately, you’ll see that the results and conclusions are not changed at all when using all 11 bubbles, or just the big 8 ones.

Table 1: List of bubbles used to analyze price characteristics around a bubble peak.


Figure 1 shows each price move and a composite of the average price return for all bubbles (green circles) spanning a time period of three years before the peak (~756 trading days) to three years after the peak. Day 0 corresponds to the peak price for all 11 bubbles. Prices for each of the curves are referenced to this value, normalized to 100% on day 0. Note that a rise from 10% to 100% corresponds to a 10-bagger, while a rise from 50% to 100% is a double. Weekend pricing was eliminated on some indices (e.g., bitcoin, Nikkei 225 and DJIA) to approximate 252 trading days in one calendar year.

Figure 1: Price characteristics of 11 bubbles over a six-year period centered on the market peak date.


What can we learn from this chart? First, we see that the rate of price appreciation accelerates 1 to 1.5 years before the bubble peak. The excitement associated with everyone getting rich, the price-insensitive buying by retail investors, and difficulties associated with shorting leads to a blow-off top. No one knows when the last buyer will buy, just as no one knows when the last seller will sell during a market panic.

Figure 2 shows the same plot with the big 8 bubbles. I’ve also drawn a trendline using the first 350 days (400 trading days before the peak), and we can see the price appreciation begin to diverge from the line. The trendline already represents an unsustainable high rate of return of roughly 18% per year.

We can think of the trendline as one estimate of fair value. The bubble dynamics force the price well above fair value. This curve looks like an upside-down version of a panicky market crash, or on a shorter time scale, a motivated selling event, such as forced selling due to a margin call. In the latter case, non-economic selling temporarily pushes prices below fair value due to the lack of arbitrageurs to jump in and rationalize the price. A bubble is the same thing to the upside, but much larger and longer in duration. Of course, the efficient market guys would suggest that fair value is equal to the price at all moments during this boom and bust sequence. Probably fair value is somewhere between the trendline and price, but I’d guess closer to the trendline than the price.

Figure 2. Price characteristics of the big 8 bubbles over a six-year period centered on the market peak date.


Once the bubble pops, prices fall back to fair value, which for Figure 2 is still 50% higher than it was three years earlier when the price diverged from the trendline. Even if much of the new-era thinking is correct, prices just got way ahead of themselves when incorporating the new highly positive information.

To gain a further view of how extreme these market moves are, we next examine every non-bubble, cyclical, bull-market top for the U.S. stock market going back to 1916. Our data series chain-links the Dow Jones Index (DJIA) data to S&P 500 Index data to finally SPY total return daily price data. Bull-market top dates were taken from Ned Davis Research Table T_200B.RPT, which uses mechanical rules to define bear markets. The primary rule is that a bear market is defined by a 30% drop in the DJIA after 50 calendar days, or a 13% decline after 145 calendar days. That’s the total of 32 bear markets since 1916.

Some of these bear markets are associated with the 1929 and 2000 stock market bubbles. I’ve eliminated three cyclical bear markets associated with the 1929 stock market bubble and two cyclical bear markets associated with the 2000 stock market bubble. Most people tend to think of the declines associated with these bubbles as one big bear market, rather than a series of multiple small bear markets.

Figure 3 shows a composite of the 27 “non-bubble” bear markets, along with all the individual bear market price curves. Many of these non-bubble markets were quite severe due to World War I and World War II (1916, 1919 and 1939), a severe banking crisis (2008), an inflationary crisis (1973-1974), large economic contractions (1937) and a flash crash (1987).

Figure 3: Average of all NDR-defined, bull-market tops with those associated with the 2000 and 1929 bubbles removed (NDR bull markets peaking in 1929, 1930, 1932, 2000 and 2002).


What can we learn when comparing Figures 1 and 2 with Figure 3? First, the rate of return in the three years leading up to the top is dramatically different. The bubble composite rises about 3-6x over the three years leading to the bubble top. Prices rise parabolically into the top. To contrast, equity bull market tops are relatively mild, drawn out affairs, with tops taking their time with many head fakes. The average return over the three years leading into a top is a relatively large 40% over three years, but not in the same league as the 6x associated with the 11 bubbles or 3x associated with the big 8 bubbles.

Typical stock market tops are impossible to predict. However, the accelerating nature of the price rise, along with the presence of bubble attributes shown above, and the large divergence from the trendline, suggest that a bear market top associated with a bubble is a bit more predictable (within a year or so).

Also, a typical bear market bottoms in about a year, then heads off to new highs. Prices make new highs 1.5 to 2 years after the previous peak. For bubbles, that time frame can be much longer (as might be expected). Figures 1 and 2 suggest that a bottom is put in between 2 and 3 years after the peak, but often the effective low is out much longer (as we’ll discuss later). With respect to making a new high, that can be a decade away after a bubble.

Table 2 shows the median annualized return for a variety of time frames before and after the bubble peak for the 11 bubbles, the big 8 bubbles and the 27 non-bubble, equity bull-market tops. Rather than averaging returns, as we did in Figures 1-3, we are now tabulating median returns, which helps eliminate the impact of outliers (bitcoin).

Table 2: Median annualized returns for various time frames around bubble and stock market peaks.


Note that for bubbles, the median annualized returns accelerate as we get closer to the peak. Within six months, the median annualized return is 174% among the 11 bubbles and 165% for the big 8, which is much greater than equities near the top of a typical bull market. The median return during the last year of a bull market is 20% per year, which is twice the long-term return of the stock market, but not in the same league as the “get rich quick” returns associated with the last year of a bubble.

For all time frames beyond the peak, median returns for the bubbles are all in the negative territory, except for one case associated with the third year of the 11 bubbles. For non-bubble bear markets, prices are beginning to move higher by the start of the second year beyond the peak, and annualized returns have returned to normal levels of about 10% per year.

What about volatility? Table 3 compares the median annualized volatility over the same time frames. For typical bull-market tops, volatility is around normal levels of 14%. Volatilities actually drop leading into the last six months before the top. After the peak, volatility picks up, as is usually the case when equities fall.

Table 3: Median annualized price volatility for various time frames around bubble and stock market peaks.


Figures 4 and 5 show the trailing three-month volatility associated with the big 8 bubbles and the 27 non-bubble, bull-market tops.

Bubble characteristics are from a different planet! Volatility levels associated with bubbles are extremely high to begin with (in the 30% range before the peak), and tend to rise as the bubble peak gets closer. After the bubble pops, volatility levels spike to extreme levels with a median volatility in the 50% range, which gradually falls over time. Volatility levels this high can play havoc with trend-following systems with whipsaws.

Why is volatility so high? It’s anyone’s guess. Perhaps there are very few buyers left to buy, which can lead to some serious air pockets to the downside. Or short-term traders, who most likely played the trend to the long-side, begin to rack up losses by manically chasing short-term price moves to either cut losses or not miss the next big upside move.

Figure 4: Trailing three-month volatility (66-day) for the big 8 bubbles (note the log scale).


Figure 5: Trailing three-month volatility (66-day) for the 27 non-bubble, bull-market tops (note the log scale).


Mechanical System Trading Approaches

We can use these daily data series to investigate various mechanical trading approaches while taking care to account for stale pricing effects associated with backtesting index data. I won’t discuss the details, but I’ll talk briefly about expectations.

The bubble composite curves in Figures 1 and 2 seem to indicate that a long-short, trend-following system could work very well on the way up and also during the downside of a bubble. This is indeed the case before the bubble peak since practically all trend following models are long the market without whipsaws in an environment of parabolic rising prices. But what’s interesting is that the volatility of each individual bubble is so high that whipsaws destroy returns associated with shorting the market after the peak with trend following. Post-bubble peak, intermediate trend-following systems are still long the market 40% of the time.

Post-bubble, it’s best to avoid using trend following if possible; although, another option is to dramatically increase the time scale associated with the model.

A better approach might be to employ a countertrend model after the bubble peak on the short side to exploit the extreme volatility. Any sort of sell-strength/cover-on-weakness model can work. With this approach, you’re exploiting a price that is likely way above fair value and a price volatility that is extremely high.

Long-Term Bubble Characteristics

There are a few questions about the long-term nature of bubble bursting events. First, what is the time scale for the ultimate bottoming of the bubble asset class? Even if prices have bottomed, is it a great buy, or does the asset class continue to be “dead money,” losing value versus other asset classes and inflation for much longer time periods?

Second, does the bursting of a bubble cause economic recessions? Or, if a bubble is small compared to the size of the economy, does it provide an indication of excessive optimism associated with economic and stock market tops?

Finally, does the popping of a bubble create a catalyst for major asset class trend changes? Usually the bubble asset class has enjoyed a decade of fabulous returns. For the next few years, the bubble asset class will be losing value. Does investor attention shift to a new, previously ignored asset class? Perhaps this process triggers a flow of funds from the bubble asset class to other more attractive investments, and these flows represent a headwind for the bubble asset class.


Let’s start simple and examine the collectibles bubbles shown in Table 4. Along with peak date, the table also shows the duration of the decline, magnitude of the decline (after inflation) and recovery time. I’ve added two more bubbles to the list since the previous Bubble blog post – stamps in 1980 and art in 1990. I haven’t fully vetted these bubbles versus the bubble attribute list, but they appear to meet enough of the 10 bubble attributes to be included in the table.

With respect to their size versus GDP, these bubbles were quite small compared to equity and real-estate bubbles. We shouldn’t expect the popping of these bubbles to cause a recession or create powerful, multiyear outflows to fuel price rises of another asset class.

However, a collectibles bubble may provide a “canary-in-the-coal-mine” tell that speculative juices are flowing in all markets, consumer and investor optimism is high, and perhaps all good news is priced into risk assets. British stamps peaked when other inflationary assets peaked in 1980. Rare coins peaked a year before the 1990 recession and equity bear market.

Table 4: Long-term performance of collectibles post-bubble peak.


The 1990 art-market peak coincided with the bubble peak in Japanese stocks and the 1990 equity bear market. I included two art indexes where I could find data associated with this bubble. The Japanese were very active in the art market, particularly with Impressionist paintings. It’s reasonable to expect that different art indexes were affected quite differently, depending on their composition.

Baseball cards peaked during the 1994 mini bear market. Beanie Babies peaked a year before the 2000 stock market bubble and bear market of 2000 to 2002. This small sample does suggest that if you notice a bubble occuring in another market, no matter how small, it could be a harbinger of an upcoming equity bear market.

The next thing to notice is that the duration of the declines (after inflation) is long, with a number of these bubbles still not seeing a bottom in pricing after 20+ years of declines. There’s no doubt that post-bubble peak, these assets are dead money for a long, long time.

On a final note, one lesson I learned from studying these collectibles bubbles is the importance of rarity. For collectibles bubbles, it’s the low-priced, easily affordable items that the general public bid up to crazy prices, and then take decades to recover. This is often the market segment that experiences the extreme bubble activity and the highest percentage gain as the bubble grows.

The extremely rare and most coveted items, which are only affordable to the ultrarich, often do not experience the same rate of return on the way up. Moreover, the most valuable items don’t fall as much post bubble and eventually recover and soar to new highs. Compare the historical charts of the PCGS3000 Index (much more affordable coins) to the PCGS Rarities Index and the PCGS Ultrararities Index.

This effect is observed time and again. The most coveted baseball card, the T206 Honus Wagner, was not affected by the bursting of the baseball card bubble and is now trading at prices much higher than in 1994. The rarest stamp in the world, the British Guiana 1c magenta, also weathered the British stamp bubble better than the average stamp. Most Beanie Babies lost all their bubble peak value, but there are some rare ones, such as Peanut the Royal Blue Elephant, which is being offered at much higher prices than the original $5 retail price.

The bottom line is that the low-priced stuff that was affordable for most new collectors will likely be dead money for at least a generation after the bubble peak. The extremely rare items will eventually recover, although they may be dead money versus other asset classes for a decade or more.

Stocks, Real Estate and Commodities

Table 5 shows the longer-term performance for a variety of bubbles throughout history (post-bubble peak). This data is often difficult to track down because it seems that people are most interested in tales of bubble euphoria and the subsequent crash. Apparently, the longer-term implications are less interesting  .

Table 5: Long-term, post-peak characteristics of historical bubbles.


The above table shows post-bubble, price-decline characteristics using total return data (when available) for stocks and price data for real estate and commodities. It shows the duration of the decline in years, the magnitude of the decline and the recovery time for the asset class to reach a new nominal high since the time of the bubble peak. The magnitudes of the declines are very large, and interestingly, many bubbles bottom many years beyond what’s implied in the bubble charts in Figures 1 and 2.

The statistics in Table 5 range from depressingly long periods of decline (Japan, gold and silver) to relatively short declines of a couple years, indicating that perhaps a great buying opportunity was at hand near the bottom (stocks and bitcoin).

Are bubble assets great buying opportunities when they finally bottom? To examine this question, I compared the nominal returns with the opportunity costs. For real estate and commodities, I used inflation as a measure of opportunity cost. For stock markets, I used the MSCI All-Country World Index (ACWI), or the S&P 500 for the energy stock bubble of 1980.

When compared to the opportunity costs, the damage that bubbles do to future returns is much worse. The length of declines is five years or longer for practically all bubbles. Out of this list, there were only two great buying opportunities when the market bottomed. The first was after the 1929 crash, which caused stocks to lose 83% of their value (by far the worst drawdown in U.S. stock market history). Interestingly, catching the bottom was difficult because stocks bounced by about 100% off the bottom in just two months. The other case was the 2013 bitcoin bubble, which as we now know rocketed back to new highs about a year after the bottom.

While NASDAQ and S&P 500 prices bottomed after 2 to 2.5 years of declines, these assets underperformed for another four to six years, while other equity asset classes outperformed, such as emerging market stocks, value stocks and REITs. Chinese equities bottomed just 1.2 years after the 2007 top, actually before the rest of the market bottomed in early 2009. But Chinese stocks, along with emerging market stocks, were serious laggards versus the ACWI. U.S. stocks became the winners for the next decade following the 2008/2009 crisis.

Of course, the 1980 bubbles signified the end of the secular bull market of inflation and inflation-hedging assets and the start of new secular bull market in equities and bonds. Energy stocks went from 30% of the S&P 500 in 1980 to just 5% in late 1999. Commodity bubble assets, Melbourne and Japanese real estate, and collectibles were dead money for decades after the bubble peak.

As a general rule, the lesson is that asset class bubbles lead to underperformance for a minimum of five years, and perhaps much longer (even decades). It appears that once burned, investors have no desire to risk losing a lot of money again in the same asset class. Bitcoin in 2013 is the glaring exception to that rule.

Do Asset Price Bubbles Cause Economic Crises?

When the bubble is really small, such as with the collectibles markets or with bitcoin in 2013, there should be no economic impact when the bubble bursts. As discussed earlier, the formation of these “micro-bubbles” may provide an indication that the good times have gone on far enough, and that an economic downturn and/or equity bear market is on the horizon. Remember that one of the bubble attributes is that they occur late in the economic cycle after many years of prosperity leading into the final parabolic runup.

When the bubble is large, on the order of a nation’s gross domestic product (GDP), then we should expect economic consequences when the bubble bursts. This is the basic message of Table 6, which shows the estimated size of the market at the bubble peak compared to a relevant GDP measure and the post-bubble economic impacts.

Table 6: Summary of economic impacts of bubbles.


Real-estate bubbles can seriously impact banks, such that falling prices can actually influence bank lending, which can in turn affect the economy. This is one of those times when prices affect fundamentals, rather than reflect fundamentals.

The 1980 bubbles were different – they were driven by fear associated with runaway inflation. It was a highly uncertain time, with central banks aggressively raising short rates, forcing the economy into a recession.

Also, a large bubble (compared to GDP) can serve to fuel bull markets elsewhere as investors slowly give up on the asset class over the next few years. This certainly happened when the NASDAQ and S&P 500 peaked in 2000.

The bottom line appears to be that big bubbles (as a percent of GDP) typically lead to an economic crisis and/or contraction when the bubble finally bursts.

On the Way Up: It’s a Bubble, so Now What?

If we make the educated guess that bubble mania has taken over asset pricing, how can we take advantage of this? Almost by its very nature, the setup is difficult. We may feel strongly that prices deserve to be much lower, yet we don’t know when that will occur, or how much pain we need to endure to get there. The top won’t come until all buyers are exhausted, including capitulating shorts and reluctant investors finally throwing in the FOMO towel.

Shorting a frenzy can at best gain 100%, while possibly losing multiples of that. Implementing a short-side trading approach doesn’t improve the reward-risk ratio much – the expected gain compared to the potential loss remains unattractive. There are just so many ways for things to go wrong with any shorting strategy that a trader can end up losing money, even if they’re ultimately right when prices finally collapse. This is one reason why the smart money simply sells and steps away in these situations.

It’s interesting that when we hear the stories associated with past bubbles, we don’t hear much about anyone getting rich from the collapse in prices. We just hear about the people who rode the wave higher and sold, those who rode the wave up and then back down, or those who lost everything buying near the peak.

Michael Lewis’ The Big Short26 is a great book that chronicles a few traders who profited immensely during the mid-2000s housing collapse. These investors – and there was just a handful of them – endured immense pain in this trade, even though it was uniquely asymmetric in their favor. As the story unfolded, if I were in their shoes, I know I would have given up on the trade well before the large payoff due to a variety of risks.

Of course, another idea is to make the bet that the bubble peak is at least six months away, and implement a long-only, trend-following model to protect on the downside. If the peak is indeed months away, then you’ll likely come out ahead with this strategy.

The Bubble Popped, so Now What?

If you make the educated guess that the bubble has finally popped, perhaps three to four months after the peak, it will be tempting to short the bubble. Every bubble is different, with its own features and constraints. Emotions will drive prices, and the heavy retail investor presence should present a few asymmetric trade opportunities on the way up and down. You can’t use trend following on the downside; there’s too much volatility.

Here are some ideas:

  • Implement a countertrend trading strategy. Short rallies, then cover dips. Do this on any time scale. This should work for at least the first year after the bubble has popped due to the excessive volatility and a price that’s way above fair value.
  • Short the weakest links. Every bubble has the crappy product that was sold to retail investors during the craze, but will probably be worth nothing a few years after the peak. Short the low-priced vehicles, short closed-end funds and ETFs with large premiums, and short stocks that are related to the theme in name only. Unfortunately, these trades are probably crowded, expensive (with high borrow rates), and sensitive to short-squeezes.
  • Search for relative price anomalies. Find pair trades where the payoff is asymmetric and less sensitive to the price of the bubble asset. Use your imagination to search for opportunities. Perhaps there are closed-end funds trading at large premiums or discounts. In his book Irrational Exuberance, Robert Shiller comments on the crazy valuation of eToys versus Toys “R” Us or irrational pricing of Palm versus 3Com during the 2000 NASDAQ bubble.
  • Short the market at an appropriate size level. Then stick with the trade with high conviction, because it will be tested by many strong countertrend rallies.
  • Search for tells and/or catalysts that the bubble has indeed popped. A catalyst may be a major hedge fund capitulating on shorts. Another catalyst is the introduction of more trading vehicles (such as futures contracts) that enable shorting the bubble. Look for tells such as bad news no longer being ignored, or notice when the good news is being ignored. What about new product introductions? On the way up, prices of these new products soar upon their introduction. Watch for when this no longer happens. Watch for signs of desperation among the “bubble geniuses” to prop up prices after the peak. All of the catalysts and tells may help make shorting more profitable.

Best Approach – Avoid the Bubble

Perhaps the best approach is to sell anything related to the bubble, and just avoid trading or investing in the bubble asset. Place it in Munger’s and Buffett’s “too difficult” box and search for trading and investing opportunities elsewhere. As an investor, this is by far the best approach. This is also an excellent approach for an asset class trader.

It appears that any bubble is a signal for an upcoming bear market and economic downturn. If the bubble is large compared to an appropriate measure of GDP, then we can expect a serious downturn. It may pay to get a little cautious during these moments as a bubble grows. Be open minded about the possibility that a bubble anywhere in the world may be a sign that there’s a lot of optimism priced into risk assets. This certainly seems to be the case for the current cryptocurrency bubble. There was a lot of optimism priced into risk assets at the beginning of 2018.

This is a new lesson for me – to be open to the idea that any bubble can be a signal for future weakness in risk assets. It’s interesting that the mini bear market associated with the ACWI in 2015/2016 exactly traces out the path of the China A-shares bubble bursting. We sold our A-shares ETFs near the top in 2015, correctly viewing the rise as a retail-driven bubble. But I had no idea that the popping of this bubble, which seemed to affect only China, would be a harbinger for a worldwide correction in risk assets. It didn’t cause the downturn, and certainly no one attributes the ACWI correction to the bursting of the A-shares bubble. But it was a signal of potential weakness.

When the bubble finally bursts, look for other assets to begin outperforming as investors shift their attention elsewhere. The bubble bursting serves as a catalyst for new asset classes to become star performers. Obviously, when the NASDAQ bubble popped, all sorts of non-tech assets became the place to be for the next six to eight years. The good bet is that the bubble asset will be dead money for at least five years after the peak. There will likely be years of selling pressure hanging over the asset class as investors slowly give up on it.

Also keep in mind that you may have to wait for the end of a bear market to put these trades on. Another lesson for me is that I ignored the potential for a major secular asset class change after the China and oil price bubbles of 2007 and 2008. Coming out of the 2009 bear market, I was still positive on emerging market stocks and negative on U.S. stocks. I should have realized that the China bubble popping would trigger new asset class outperformance trends.

Eventually, after a half a decade’s time, the high-quality assets associated with the bubble asset class may be trading at bargain prices. That may be the ultimate way to trade the bubble asset – when no one cares about it anymore.

In-Benchmark Bubble

Career risk can be a major factor that inhibits professional portfolio managers (PMs) from doing their part to push bubble prices back to fair value. As fast-rising prices get more “bubbly,” selling the asset class into cash might anger your investors, who are likely caught up in the craze. They may immediately fire a PM who’s not providing full exposure to such a “great investment opportunity”.

PMs who were in the U.S. growth equity style box in 2000, Chinese equity in 2007, or Japanese equity in 1990 all faced this conundrum. What can such a PM do? Any approach to generating alpha associated with the bubble will also come with career risk if the implementation and timing is wrong. One approach is to wait for the obvious bubble peak, and then shift the portfolio into other risky assets to avoid multiple years of underperformance to follow. A PM can also try to invest in the highest quality names, which should outperform on the downside, but underperform on the upside. This sort of career risk trade-off is why bubbles can tie the hands of many professional PMs while prices diverge from fair value.


I write these blog posts to reflect and get better. What have I learned?

I think it’s possible to distinguish a bubble from an “appropriate” large price move using the 10 bubble attributes described in my previous post. When a bubble hits that parabolic stage, the peak is likely within a year or so.

Bubbles are frustrating animals. It’s not every day that you can make an educated guess that prices have diverged dramatically from fair value. But exploiting the opportunity is tougher than it seems. Short-sale costs and constraints often reduce the effectiveness of shorting. It’s interesting that trend following on the short side is thwarted by the extreme volatility associated with the post-bubble price action.

Probably the best approach is to avoid the bubble asset class and search other markets for attractive investments. Be open to the idea that any bubble, large or small, can be a harbinger of an economic downturn and bear market. Obviously, the larger the bubble compared to the economy, the more likely there will be negative consequences when the bubble bursts.

Be open to the idea that the bubble asset class will be dead money for years to come, maybe even decades. Search for the new asset class emerging as the outperformer when the bubble finally pops.

Finally, what are the implications of the cryptocurrency bubble? It appears that the January 7, 2018, high is the final bubble peak in cryptocurrency assets at a market cap of $830 billion. Bitcoin peaked earlier on December 17, 2017. The total cryptocurrency market cap has fallen almost 70% as of March 31, 2018. I’d bet that the $830 billion market cap high will not be revisited for a long time and that crypto-assets will be dead money for many years.

The current cryptocurrency bubble is relatively small in size, on the order of 1% of the world’s GDP at its peak in January. We don’t expect this bubble bursting to have a major impact on economies around the world, yet it may be providing a clue that it’s time for a bear market. Stocks had a great run in 2017 and enjoyed immense gains since the bottom in March 2009. The collapse of another bubbly asset, the short-volatility funds (for example, see the stock ticker SVXY) on February 5, 2018, may also be another canary tweeting “turbulence ahead.”

Going into 2018, positive expectations for stocks and the economy were excessively high. There were few attractively priced risk assets. At the time of this writing, stocks are in correction mode, and only time will tell if the cryptocurrency bubble has signaled the next bear market.



  1. Greenwood, R., Shleifer, A., You, Y., “Bubbles for Fama”, Working Paper, February 2017.
  2. Chancellor, E., Devil Take the Hindmost, 1999.
  3. Dimson, E., Spaenjers, C., “Ex post: The investment performance of collectible stamps”, Journal of Financial Economics, Vol. 100, No. 2, 2011, pp. 443-458.
  6. Neuendort, H., “Academics Say the Art Market Bubble is About to Burst – Are They Right?”, artnet news, January 19, 2016.
  7. Goetzmann, W.N., Renneboog, L., Spaenjers, C., “Art and Money”, American Economic Review: Papers & Proceedings 2011, 101:3, pp. 222-226. See also 2010 version of the working paper.
  8. Hwang, J., The Motley Fool, February 25, 2013.
  9. Bissonnette, Z. The Great Beanie Baby Bubble: Mass Delusion and the Dark Side of Cute, 2015.
  10. Mahtani, S., “Sorry, Collectors, Nobody Wants Your Beanie Babies Anymore”, Wall Street Journal, February 21, 2018
  11. Frehen, R.G.P., Goetzmann, W.N., Rouwenhorst, K.G., “New Evidence on the First Financial Bubble”, Yale International Center for Finance Working Paper No. 09-04, July 27, 2012.
  12. Hoyt, H., One Hundred Years of Land Values in Chicago, 1933.
  13. Simon, J., “Three Australian Asset-price Bubbles”, Reserve Bank of Australia, 2003.
  14. Acheson, G.G., Hickson, C.R., Turner, J.D., Ye, Q., “Rule Britannia!: British Stock Market Returns, 1825-1870”, The Journal of Economic History, Vol. 69, No. 4, December 2009, pp. 1107-1137.
  15. Guinn, J.M., Beck, J., A History of California and an Extended History of Los Angeles, 2015, chapter 38.
  16. Stapledon, N., “A History of Housing Prices in Australia 1880-2010”, The University of New South Wales, School of Economics Discussion Paper: 2010/18, September 2010.
  17. Turner, G.M., The Florida Land Boom of the 1920s, 2015.
  19. Many potential data sources, such as
  20. Colombo, J., Kuwait’s Souk al-Manakh Stock Bubble, May 12, 2006. See also:
  21. Morningstar Advisory Workstation
  22. Wood, C., The Bubble Economy: Japan’s Extraordinary Speculative Boom of the ‘80s and the Dramatic Bust of the ‘90s, 2005.
  24. S&P/Case-Shiller U.S. National Home Price Index,
  26. Lewis, M., The Big Short: Inside the Doomsday Machine, 2011.
  27. Shiller, R.J., Irrational Exuberance, Second Edition, 2005.




The content contained within this blog reflects the personal views and opinions of Dennis Tilley, and not necessarily those of Merriman Wealth Management, LLC. This website is for educational and/or entertainment purposes only. Use this information at your own risk, and the content should not be considered legal, tax or investment advice. The views contained in this blog may change at any time without notice, and may be inappropriate for an individual’s investment portfolio. There is no guarantee that securities and/or the techniques mentioned in this blog will make money or enhance risk-adjusted returns. The information contained in this blog may use views, estimates, assumptions, facts and information from other sources that are believed to be accurate and reliable as of the date of each blog entry. The content provided within this blog is the property of Dennis Tilley & Merriman Wealth Management, LLC (“Merriman”). For more details, see the Important Disclosure.