Bubble Testing Proves Prices Can Be Irrational

In a previous article, I offered details on one way to quantify a bubble, at least for defining a market opinion. In this article, we will test the idea and assess the probability of successfully trading bubbles. As a result of the tests, we can develop some insights into markets and we’ll conclude with a successful trading strategy.

It is popular to assume a market is in a bubble when prices have moved too far, too fast. This definition ignores the question of value and scarceness. Sometimes prices rise because they should – the stock price of Hansens Soda increased more than 18,000 percent in less than ten years because sales and earnings rose. Value moved the price. A lack of scarceness, in other words an ample supply, also means prices can increase sharply without being in a bubble. It seems impossible for Treasury securities to be in a bubble regardless of price because there is always an impending increase of supply. With the federal government debt growing at $1 trillion a year, the new supply prevents a bubble.

Gauging value and supply, necessary preconditions of a bubble, requires fundamental analysis. For home prices, it could be done by looking at the price-to-rent ratio or a housing affordability index. Supply is measured by the number of listings in a real estate market. Value and supply can be quantified in almost any market. Traders can ignore these factors and focus solely on price since they should be more interested in profiting from trends and less interested in underlying causes.

Bollinger Bands® can be applied to the annual rate of change to help identify an excessive price move. Bollinger Bands quantify when price moves have become excessive and are commonly applied to prices. This is a very useful indicator that can be applied to any indicator. When added to the rate of change indicator, Bollinger Bands can be used to visually spot a price move that is over extended. Theoretically, the ROC should be outside standard Bollinger Bands about 5 percent of the time. Logically and visually, this is an interesting idea.

Some traders can make money using only chart patterns. Many more traders will discover that patterns confirm what they want to see and chart trading all too often leads to losses.

Bollinger Bands on an indicator is an easily testable idea. We can assume that prices are in a bubble when the ROC is above the upper Band and test the idea of selling when the ROC falls back under the upper Band line. The trade set up will occur when prices are showing an unusually fast price surge. We should expect this set up to identify about 2.5 percent of the time in any market. The trigger for the trade will be when the rate that the price is changing falls back to a normal level. In theory, we are getting short when the bubble bursts.

On the S&P 500 index, using all available data and exiting after a one month holding period, this strategy worked 53 percent of the time and would have made money. The opposite idea, going long when the ROC is more than two standard deviations below average only identified winning trades 43 percent of the time. The results are even less promising for a shorter holding period. None of the test results are good enough to rely on as a trading strategy.

Results are slightly worse when the Bollinger Bands are widened to 3 standard deviations. So far, the test results seem to be confirming that an overbought market can become more overbought and it is actually a sign of strength when a market moves rapidly.

An interesting insight can be made when the Band width is set to 6 standard deviations. At this setting, the Bands should contain 99.99966 percent of the data, meaning we should see three moves beyond that level in a million trading days. Instead we have seen 41 trade signals using this indicator in only 28 years, a time frame including about 7,000 trading days. Only two of these extreme price moves were up side moves, a fact that demonstrates the stock market tends to have extreme sell-offs more frequently than price bubbles.

One accepted characteristic of bubbles is that they result from the irrational behavior of traders, and it was probably too much to ask for a simple and rational indicator that could be used to consistently spot the end of a bubble.

Noting the poor test results, we can flip the idea. An accelerating ROC that breaks above the upper Bollinger Band could be a good buy signal. This leads to a strategy that identifies winning trades 60 percent of the time with a one month holding period. Risk, measured by the size of the draw down, is too high for this to be a tradable system. Shorter holding periods all lead to losing systems.

Although there are very few trades, using 4 standard deviations or more with this strategy is very successful. That means bubble-like stock market increases are the best time to buy for short-term trades, “best” meaning a chance for quick, high probability profits.

Applied to a basket of commodities, the results are similar. This is an interesting indicator, but bubble spotting is not proving to be a tradable idea.

A similar strategy can be applied directly to price and the results are also disappointing. When stock prices move beyond the Bollinger Bands, it is neither a sign of tradable strength or weakness. No consistent profits result from this action.

A key insight to draw from this result is that successful systems need to be firmly grounded in logic. Although we can create a series of logical rules for timing bubbles, the underlying behavior of the markets is irrational during bubbles. No matter how much we try to tame the behavior of crowds, bubbles do not seem to be tradable.

Another important result from the testing is that markets become irrationally oversold more often than they become overbought. Extreme selling accounted for about 95 percent of the time that the rate of change moved more than two standard deviations from normal. This demonstrates that fear is the strongest emotion driving markets. A market consolidates the behavior of thousands, in some cases millions, of individuals. It is often said that individual decisions are driven by greed and fear. While that may be true, it appears that buying is done with at least some degree of thought but sell offs occur with greater intensity.

One last insight can also be derived from this exercise. Markets aren’t as simple as they seem, and they aren’t as complex as most traders think they are.

These tests began as an effort to find bubbles. Knowing a market is in a bubble would provide a trader with bragging rights and it would lead to a degree of comfort in trading. From a trading perspective, the initial premise was flawed. The motivation behind the testing was to prove that markets could be understood. For winning systems, the initial motivation should be to find a winning strategy. Isolating that effort to less than 5 percent of the market action is too precise of a target for successful trading.

Trading requires the acceptance of uncertainty. We can’t look to the most irrational moments as a chance to profit. Success requires a coherent strategy designed to take advantage of market opportunities in any environment.

To that end, a surprisingly simple tool is the MACD indicator (using default settings) on a monthly chart. In the long-term, the system is right on 100 percent of its signals in the SPY (an exchange traded fund that matches the S&P 500). Draw downs are tolerable and this is a tradable system. It is accurate 80 percent of the time on the longer history available from back testing the S&P 500 futures contract.

Weekly charts introduce more uncertainty. MACD signals are correct about 40 percent of the time on the SPY trades and only a third of the time on the S&P futures contract. However, profitability is significantly increased and more than offsets the minor increases seen in the draw downs. As a stand alone trading strategy, MACD could work.

If it is more important to be right than to have an infallible market opinion to share with friends and clients, then trading is not for you. Trading success requires taking risks, and the ability to endure being wrong up to two thirds of the time. Baseball players get to the Hall of Fame with one hit every three at bats, traders can get very wealthy with the same percentage of winning trades.

Trading is about making money, not necessarily being right. Accept uncertainty and make forecasts only as a hobby.

By Michael J. Carr, CMT