Relative Strength Rotations

I was reading http://cssanalytics.wordpress.com/2010/02/26/rotation-part-3-beta-surfing/ and associated information from Market Rewind .  I decided to experiment some with the ideas.  The first experiment on relative strength was to pick the top percentage gainer over the last X months.  I found that this did not have much consistency.  I found a great deal of variation depending on the day of the month (last, first, option expiration),and comparing the results from using a period of X weeks to X+1 weeks had drastically different returns and therefore didn’t feel it was particularly robust.  So, using the Market Rewind blog post, I decided to experiment with defining relative strength as a comparison of current price vs a simple moving average.  I varied the length of the moving average and was hoping to find that the results weren’t much different when comparing a moving average of X weeks and X + 1 weeks, and that is what I found.  I decided to use a 26 week moving average.

In the spirit of the CSS analytics post, I choose for stocks: SPY, EEM, SHY, short SPY (I was going to use SH, but it didn’t have quite the history), FXE, and DBC.  The algorithm was to choose the stock which has the highest ratio of price / SMA(26 weeks), using Friday’s close.  Hold for 1 week,  don’t include commission etc. I am using Yahoo historical data, adjusted close.   If the stock didn’t exist during the week, it wasn’t included.  Starting in early 2003 and rolling forward, the gain was about 600%.

How often was each stock selected:

SPY 45
EEM 206
SHY 7
short SPY 62
FXE 11
DBC 39

I then did an analysis of how each stock performed compared to the others. For this table, each row is when the stock in column 1 had the best ratio.  Each column is the average weekly price change. I am using the average change in price for the next week, which is not as good as other calculations, (such as CAGR or even median), but it was easy to create the table in Excel.  For these results, I used data going back to late 2006 – DBC started trading in 2006.

top class\average weekly performance SPY EEM SHY short SPY FXE DBC
SPY 0.73% 0.24% -0.06% -0.73% -0.46% 0.32%
EEM 0.28% 0.73% 0.08% -0.21% 0.22% 0.45%
SHY 1.04% 2.55% -0.03% -1.04% -0.15% 1.12%
short SPY -0.96% -1.01% 0.11% 0.85% -0.42% -1.69%
FXE 1.38% 0.56% 0.20% -1.37% -0.72% -0.14%
DBC -0.21% -0.12% 0.07% 0.31% 0.23% 0.76%

In other words, if the top stock was SPY, then on average, it gained .73% the next week (top left corner), and on average, the best performer when SPY was the top stock. Similar results are for EEM, short SPY, and DBC.  On the other hand, for SHY, when it was the top performer, EEM had better returns the next week, and similar for FXE, where SPY was a better performer.  These two didn’t have a large number of times when they were the best, so the statistical significance is less than the other stocks.  Since 2003, there were about 37 trades.

Advertisements

2 Responses to “Relative Strength Rotations”

  1. jt Says:

    tks for the research idea. would you be willing to share the trades spreadsheet – i must be doing something wrong since i can’t recreate the results?

  2. David Hall Says:

    Hi, I am new to systematic trading too, is it possible to share your spreadsheet?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


%d bloggers like this: