Seeking Alpha
Long/short equity, event-driven, homebuilders, currencies
Profile| Send Message|
( followers)

Summary

  • 'Buy in July' did not materialize this year, making the specter of August through October loom even more ominously.
  • August through October are S&P 500's most dangerous of the year because they deliver the largest drawdowns on average.
  • The relationship between an S&P 500 loss in July and subsequent drawdowns is strongest for August and September.

Last month, I examined the data behind the surprising seasonal strength in July for the S&P 500 (NYSEARCA:SPY). The piece was called "'Sell In May' Fans Should Brace For 'Buy In July'." The odds favored the bulls for July and trading in the month got off to a promising start with fresh all-time highs and a first-of-the-month gain. The follow-through was weak and the S&P 500 twice lost all its gains for the month. As late as July 30th, the S&P 500 looked good to cling to a marginal gain for the month. Instead, July ended with a loss as the market "broke down" on the last day of the month with a 2% plunge.

The S&P 500 closed below its 50-day moving average and its lower-Bollinger Band for the second day in a row - a sign of on-going commitment from sellers. Two months of gains have been reversed in a few days of trading.

Source: FreeStockCharts.com

My demonstration of the bullish case for July was actually missing one more piece of relevant historical data. In 2014, 5 of the first 6 months of the year delivered gains on the S&P 500. Such a display of price strength has only occurred 14 times since 1950. July delivered a loss in only five of those years - 2014 makes the record 6 of 15 times. (All S&P 500 price data are from Yahoo! Finance).

The S&P 500′s July Price Change After Trading Gains in 5 of the First 6 Months of the Year (Since 1950)

When the stock market defies a historic pattern so definitively, it is sometimes useful to examine potential implications. July is particularly intriguing because its typical strength is followed by three months that have historically caused the market the most trouble: August and September tend to be the weakest months of the year and all three months can deliver particularly large sell-offs.

The last piece on S&P 500 correlations looked at monthly performance based on rank for the year. This time, I decided to look at seasonal performance in terms of maximum drawdowns in each month. A drawdown is a loss in capital. This measure is particularly important for bulls assessing risk and bears sizing up opportunity. The chart below shows the maximum drawdown on an average monthly basis from February, 1950 through July, 2014. The maximum drawdown for a given month is where the month-to-date price change is at its lowest point for the month, based on closing prices. A positive maximum drawdown occurs when no day of the month closes below the close of the last trading day of the previous month; these are months where traders should love being long and hate being short. I calculate the average across all years for a given month. August through October clearly stand out as the most "dangerous" months of the year.

Maxmimum Drawdowns on the S&P 500 On An Average Monthly Basis (Feb, 1950 to July, 2014)

The next question then is, do these most dangerous months of the year tend to experience worse maximum drawdowns when July performs poorly? The answer is an admittedly unsatisfying "perhaps."

The simplest way to describe the change is to break out July's performance into ranges (or bins) and see what happens to the monthly average maximum drawdown. In the chart below, I have broken out July's price change as either below -2% {frequency is 14}, between -2% (inclusive) and -1% (exclusive) {frequency is 8}, between -1% (inclusive) and 0% (exclusive) {frequency is 8}, or at least 0% {frequency is 35}. For comparison purposes, I also include the overall average calculated no matter how July performed.

Maxmimum Drawdowns on the S&P 500 On An Average Monthly Basis For Aug to Oct Given July's Price Change (1950 to 2013)

The apparent signal from July's performance varies by month. It is not surprising to see particularly bad Julys signal a higher than average maximum drawdown for August and September, but it is surprising to see October get relatively "safer." This year's performance falls in the -2% to -1% bucket which suggests August will be "typical" and September and October will be slightly worse than usual. (The statistically inclined are probably protesting right now that I did not demonstrate the statistical significance of the differences across these ranges of performance. The main problem is that I am examining a population of extremes, and they are not normally distributed. I will leave it to better statisticians to explain how to create a good test of significance for these data.)

For reference, I have generated scatter plots for each month. Note the general tendency for higher maximum drawdowns in August and September the worse July performs. Within the July range of -2% to 0%, the range of maximum drawdowns is at its most extensive. October appears relatively oblivious to July's price change.

Maximum Drawdowns for AUGUST based on July's price change

Maximum Drawdowns for SEPTEMBER based on July's price change

Maximum Drawdowns for OCTOBER based on July's price change

For October, the two extreme drawdowns come from 1987 and 2008. In 1987, July printed a very serene and pleasant 4.8% gain. In 2008, the main year of the financial crisis, July only fell 1.0%. Take those two large points out, and October becomes relatively serene compared to August and September in terms of the range of possible maximum drawdowns (in other words, October's two extremes somewhat distort the averages).

This analysis of course does not demonstrate cause and effect. A reason pretty much always exists to be bearish on the market. It just seems that from August to October, market participants are most willing and ready to listen to and act upon the siren calls. In the meantime, the S&P 500 is essentially oversold (meaning selling has likely reached a short-term extreme), and I am expecting at least a short-term bounce. The history of the coming three months suggests not to expect much more than to the upside.

Be careful out there!

Disclosure: The author is long SSO. The author wrote this article themselves, and it expresses their own opinions. The author is not receiving compensation for it (other than from Seeking Alpha). The author has no business relationship with any company whose stock is mentioned in this article.

Additional disclosure: Long SSO through call options.

Source: A Volatile Cocktail: A Poor July Ahead Of The S&P 500's 'Most Dangerous' Months Of The Year