Timing the Market - Part 2

Yesterday I published a post on timing the market that was based on a monthly bar crossing below a 10 period moving average. To be totally accurate, the author mentioned that she uses a 10 period average for the crossover, but also says that other periods such as 6, 8 or more than 10 might work depending on the market. It also seems that even numbers work better than odd numbers of periods! Something to backtest another time I guess.


This time I will expand on methods outlined by Charles D. Kirkpatrick II in his book (Invest by Knowing What Stock to Buy and What Stock to Sell) and also his web site as well as outline a new method based on his findings. I spoke about this author and this book in a previous post where I mentioned his fundamental analysis method. I will not go over his stock picking process, but what I found interesting was his market timing methods - he outlines 2 of them - one simple one and the other more complex. Given his success I thought that something could be learned from him. For reference, here is an equity curve of his current portfolio:

The lime green line is the results of his picks and they don't do too badly against the broader market. But the purple line is the result of his timing rules being applied to the stock picking screen. As you can tell by the flat line, for a big portion of 2008 and 2009, he was out of the market. He did miss the mid-year rally in 2008, but was out during the crash and suffered little drawdown compared to the market. Obviously the method is not perfect as 2010 was not great, but it could also be the results of bad picks!

But back to his market timing methods:

1. As explained in a PDF linked on his web site (Portfolio Construction using Kirkpatrick Methods), sometimes the simplest is the best. He uses 2 weekly moving averages and watches for crossovers. When the 2-week SMA of the S&P500 crosses below the 14-week SMA of the same index, he goes to cash. When the 2-week average crosses above the 14-week average, he goes back in using his stock screener. For example, lets looks at these averages over the last 2 years:

Weekly chart of SPY - Click to enlarge

The red circle point out where he would have gone to cash while the green ones point to weeks where he went back to the market. Keep in mind that these are weekly candles so while there is some whipsaw, this method does keep you in the market during good rallies and keep you out during corrections! For example, this method would have kept you out of the August and September corrections this year. Once again, you might miss tops and bottoms but it does beat buying high and selling low!

2. The method that he explains in his book is a little more complicated and deals with scaling in and out. His portfolios are based on models - mostly picking stocks with low price-to-sales ratio and of high relative strength. He tracks the model value on a weekly basis and checks it against moving averages - 12-, 26- and 52-week averages. If the value of the model crosses below the 12-week average, he sells 25% of the portfolio. If the value declines below the 26-week average, he raises 25% more cash and liquidates everything once the value declines below the 52-week average. This protects him from complete wipeouts! He proceeds the opposite way to return to market - investing 25% once the value crosses over the 12-week average, another 25% when above the 26-week average and is fully invested once above the 52-week average.

Since not everybody will track a specific model, this might be harder to implement. But in effect, we all access to a pre-made model. It's called an index or in some cases, an ETF! I am not certain that Charles Kirkpatrick would approve of this, but based on the chart below it seems that it could provide the basis of a new method.

Weekly chart of SPY - Click to enlarge

In this case, the method would call for going to 25% cash when the index crosses one of the SMA lines, to 50% cash when it crosses 2 of the SMA and all in cash once below all 3 SMA. And reverse the process for going back to market, scaling back in each time the index crosses an SMA line 25%, then another 25% up to 100% in when all SMA lines have been crossed. This method would have kept you out of the worse corrections and got you back in quickly once the corrections were over, but in a cautious manner. There is some whipsaw at the tops and bottoms (and remember, this is a weekly chart going back 8 years), but in general, you are in during good rallies and out during corrections (or short)!

Now all you need is to pick winners!

Timing the Market

A while back wrote a post about a method for timing the market that I read in an old article of Active Trader (Mebane Faber - April 2009). The author claimed that using a simple moving average on a monthly chart would yield better returns over time and reduce drawdowns. The strategy between 1900 and 2008 returned 10.45% a year versus 9.21% with no timing but the big difference is the 50.31% drawdown as opposed to 83.66% without timing!


Here are some illustrations of the equity curves comparison:

Keep in mind that the vertical axis is a log scale - the difference today is between $1 million for the non-timing system and $5 million with timing!


The next graphic shows the same comparison since 1972, but also adds a curve for a margin portfolio with 2x leverage (non-IRA for example)

Once again, the vertical axis is a log scale. Clearly, the Internet bubble years between 1996 and 2001 were favorable to the non-timing system, but the subsequent crash helped the timing system recover nicely - lower drawdown do help! A leverage portfolio performs much better than its 2x leverage would indicate!


With that in mind, I though that I would refresh my charts to see where we stand. So below is the latest monthly chart with a 10 period SMA as recommended by the author. 

Click to enlarge

I have circled in green the month where the system would have put us in cash. It's not perfect as for example in mid-2004 and mid-2010, we would have been kicked out in the middle of a rally. But otherwise, the system keeps up out of big bear markets! And the last signal to get out comes at the end of last month when the August monthly bar closed below the 10 period SMA! And pretty convincingly. September did nothing to help either so this might signal the start of a correction!

So, where does that take us - let's look at the 2008/2009 correction in relation to the previous rally:

Click to enlarge

We retraced over 100% of the gain with a congestion zone around the 38.2% line. Now, let's look at where we stand now:

Click to enlarge

We have retraced to the 23.6% line so far, but broken it. The next line which proved temporary resistance (38.2%) stands at around 102 on SPY. I am not making any predictions, but this would be the most logical point of resistance. In 2010, we had a mini-corrections but the 23.6% line held:

Click to enlarge

That has not been the case this time, so we might need take this more seriously!


In my next post I will outline a timing method used by another market analyst with a good track record! His method also indicates that we should have moved to cash a while back!