Testing The Ivy-10 Portfolio Ranking Score

It’s been just about six months since I’ve had an Ivy-10 Portfolio update. In this article I want to give a performance update for the Ivy-10 portfolio, answer a reader’s question and test the robustness of the the relative strength score. If you will recall, the Ivy-10 Portfolio ranks our ETFs based on a relative strength calculation. Well, how robust is parameter? Can the portfolio withstand different methods of calculating relative strength? We’ll find out.

Performance Update

How has the Ivy-10 Portfolio being doing for 2013? The results below are through June 7th. As you can see the portfolio is under performing the market as of 2013, but this is expected given the strong bull market we are experiencing.
Click for larger image.

Click for larger image.

Expanding our view out to the last major market bottom of 2009 we can see the portfolio (chart below) performed slightly better than the benchmark right up until very recently. It’s only over the past month or so have we seen a sharp drop in the performance of the portfolio.  As of this writing, both the benchmark and the portfolio have generated a CAGR of just under 17% – nothing to complain about. The portfolio has lower volatility than the benchmark.
Ivy-10-June13-2007
In previous articles we used 2007 as the starting period for our out-of-sample data for the portfolio.  Below are the results from 2007 through June 7, 2013.
Click for a larger image.

Click for a larger image.

Ivy-10-June-2013-Performance
Overall, I would say the portfolio is performing as expected.

Reader’s Question

TJ asks…
“On the IVY 10 you use GSC and DBC. These both broad commodity tracking ETFs/ETNs so the back-test is backing doubling up on the allocation on commodities. It would be interesting to see if you remove one of these then run the back test again.”
This should be simple enough. Using ETFReplay, I created two additional portfolios where I eliminated one of the commodity tracking ETFs for each of the new portfolios. I would suspect that removing one of the commodity ETFs will have little change in the overall performance of the portfolio. The results below are from 2003 through June 7, 2013.
Commodity-Test
You can see by having the two commodity based ETFs provides a balance between just using one of the ETFs.

Testing Relative Strength Score for Robustness

Like I did with my previous post in regards to testing the parameters of the Ivy-10 Portfolio, I’m going to test another parameter we have not yet tested. That parameter is the ranking score used to determine the top three performing ETFs. If you will recall, the ranking score is computed as the sum of the equal weighting of the 20-day return and the 3-month return. The results below are from 2003 through June 7, 2013.

Ranking Score

Rank Score =  ( 20-Day Return ) *.5 + ( 3-Month Return ) *.5

I call the 20-day return portion of the ranking score the Short-Term Rank and the 3-Month portion the Long-Term Rank. Together they form the Rank Score.

Testing The Long-Term Rank

For the first test I’m going to hold the Short-Term Rank untouched. I will then vary Long-Term Rank over the monthly values 2, 3, 4, 5, 6, 12, 24, and 36. Below are the results of this test.

Long-Term Rank

Looking at the graph we can see a steady decline of total returns as we increase the Long-Term Ranking parameter. The steepest portion of the decline appears at the 12 month mark and beyond. Keeping the ranking period at or under 6-months seems to be a decent value. If you recall, we currently use a value of three and that does not look like an optimize value or outlier.

Testing The Short-Term Rank

For the second test I’m going to hold the Long-Term Rank untouched. I will then vary the Short-Term Rank over the daily values 1, 2, 5, 10, 20, and 60. Below is the results of this test.

Short-Term Rank

The linear trend line drawn on this graph is opposite of the linear trend line on the Long-Term Ranking. This is telling me we don’t want too small of a value. At 10 days or higher we arrive at numbers that might interest us. However, the value of 10 does seem to be a bit of an outlier. As the portfolio currently stands a value of 20 is used and that certainly does not look like an optimized value.

Testing A Single Rank

For the third test I’m going to do away with a combined Rank Score and use a single score.  I will vary this single Rank Score over the following monthly values 2, 3, 4, 5, 6, 12, 24, and 36. Below are the results of this test.

Single-Rank-Only

Conclusion

So what does this tell us? In general I would say that a Long-Term ranking method should be based on a value of 6-months or under and a Short-Term ranking method should be no fewer than 10 days. Is there any value to combining a long-term and short-term ranking? That’s a little more difficult to determine given the results above. Combining both a long-term and short-term ranking does seem to provide more stability when looking at the the Long-Term Ranking chart and the Single Ranking Only chart. That is, each of the different data points vary less, thus providing a smoother looking curve. Put another way, it seems having a combined Ranking Score helps eliminate variation between different values by painting a better picture of which ETF is more likely to to perform  in the future. Overall, I’m inclined to keep the ranking method as is.

Other Articles:

About the Author Jeff Swanson

Jeff is the founder of System Trader Success – a website and mission to empowering the retail trader with the proper knowledge and tools to become a profitable trader the world of quantitative/automated trading.

follow me on:
>