There is a secret trading strategy for the S&P 500 that is buried in my previous article, "How To Tell If We Are In A Bear Market." In that article, I provided an indicator that provides some assistance in determining whether one is currently in a cyclical bull or bear market. To recap, I looked at quarterly closes of the S&P 500 and applied the following criteria:
1) Two sequential higher quarterly closes indicates a cyclical bull market.
2) Two sequential lower quarterly closes indicates a cyclical bear market.
3) You remain in the indicated bull or bear market until you get 2 sequential quarterly closes in the opposite direction from the current trend.
The astute reader will have realized that any indicator can always be turned into a trading strategy or trading system, and this one is no different. The rules for this profitable trading strategy are simple:
1) Go long when you get 2 sequential higher quarterly closes.
2) Exit your long and go short when you get 2 sequential lower quarterly closes.
3) Exit your short and go long when you get the next 2 sequential higher quarterly closes.
I began my backtest with the S&P 500's quarterly close of 3/31/1950 (rather arbitrarily, as that is where my data begins), and so my first buy signal was on 9/29/1950. The backtest ends on 9/30/2011, when the strategy exits its long position. The system is currently short as of 9/30/2011, but I ignore the mark to market for the purposes of this test, since it is changing daily and will only be fixed on 12/30/2011. For the purposes of this test, I assume that the S&P 500 was easily investable with futures, ETFs, or mutual funds across the entire backtest period (which it was not). I assume one could enter and exit all trades at quarterly closing prices with no slippage. I ignore all commissions, management fees, as well as dividends or interest received or paid. This is pretty much a back of the envelope test done in a spreadsheet. Any reader with more sophisticated backtesting software is encouraged to write in with more nuanced results. Here are my results:
(Click to enlarge)
We can see from these results that the strategy (taking all long and short signals, such that we are always in the market) returned 1834% since 1950. If we just took long signals, the strategy returned 3219%; and if we took just short signals, the strategy returned -53%. A simple buy-and-hold strategy over the backtest period returned 5717% without counting dividends.
1) Shorting an index is a difficult way to make money. With this trading strategy, one is better off taking just the long signals, and using the short signals as a discipline to reduce one's market exposure going into a potential bear market. Alternatively, one can take the short signals as well, but one should take profits early, and not wait for the long signal to cover one's short.
2) Buy-and-hold is inexpensive to implement, and it makes you more money than this market timing strategy. There are, however, a few caveats. A buy-and-hold investor must be prepared to endure a series of large drawdowns (sometimes greater than 50%, as we all know from our experience of the last decade). And a buy-and-hold strategy is best advised for countries that are on the ascendancy, and over periods that begin with low valuations. I would argue that the U.S. and its markets currently fail on both criteria.
3) There is a silver lining to the trading strategy that I have presented. From the historical backtest, we can see that the greatest drawdown from any entry price on the long side was -15.7%. Of course, past performance is no guarantee, and future drawdowns could be greater, but if one is willing to take the risk, one could implement a leveraged version of the long-only strategy. Using 3x leverage, the historical return would have been 9657%, with a maximum drawdown of about 47%, handily beating the return from buy-and-hold (and with a lower maximum drawdown). Today, it is possible to implement this leveraged long-only trading strategy using the S&P E-mini futures contracts. Unfortunately, it may be necessary to wait months, or even years, for the next long signal to be triggered.