Tag Archives: AGG

All-Season ETF Portfolio: Minimum Correlation Allocation

In January I launched an “All-Season ETF Portfolio”. The portfolio began with a static allocation based on a simplistic and unleveraged interpretation of Ray Dalio and Bridgwater Associates “All-Weather” investment strategy. The proposed static allocation is below:

Name Symbol Static Allocation
Vanguard Total Stock Market VTI 18.75%
 PowerShares DB Commodity Index Tracking Fund DBC 7.25%
SPDR Gold Trust GLD 7.25%
 iShares iBoxx $ High Yield Corporate Bond Fund HYG 6.50%
iShares Emerging Markets USD Bond ETF EMB 14.50%
iShares Barclays TIPS Bond Fund TIP 20.75%
iShares Barclays 20+ Year Treasury Bond ETF TLT 12.50%
iShares Barclays Aggregate Bond Fund AGG 12.50%

The proposed allocation is not intended as a one-size-fits-all allocation model, but it does serve as a framework for further study and is based on having allocation to the four different market environments espoused by Bridgewater (full paper here):

I backtested the static allocation using ETFReplay.com – it outperformed the Vanguard 60-40 (VBINX) mutual fund since 2008 on both a total and risk-adjusted basis. The All-Season portfolio has a sharpe ratio of .67 and max drawdown of -21.5% versus a sharpe of.25 and max drawdown of -34.4% for VBINX. Positions were rebalanced annually:

Last week I introduced a basic risk-parity allocation tool for the All-Season ETF portfolio. The risk parity allocation uses the trailing 20-day volatility of the adjusted closing prices of each ETF to calculate a risk-based allocation. The risk parity allocation as of Friday’s close is below:

Name Symbol Risk Parity Weighting (unleveraged)
Vanguard Total Stock Market VTI 5.17%
 PowerShares DB Commodity Index Tracking Fund DBC 7.22%
SPDR Gold Trust GLD 4.32%
 iShares iBoxx $ High Yield Corporate Bond Fund HYG 22.44%
iShares Emerging Markets USD Bond ETF EMB 12.57%
iShares Barclays TIPS Bond Fund TIP 15.62%
iShares Barclays 20+ Year Treasury Bond ETF TLT 5.41%
iShares Barclays Aggregate Bond Fund AGG 27.25%

The All-Season ETF portfolio risk-parity allocation has performed relatively well since 2008, also outperforming VBINX by a wide margin. Risk-parity posted a sharpe ratio of .82 and max drawdown of -16.8%. Positions were rebalanced quarterly:

When researching the All-Weather portfolio I came across David Varadi’s work at CSS Analytics. Varadi, along with Michael Kapler, Corey Rittenhouse, and Henry Bee,  published a paper in September 2012, “The Minimum Correlation Algorithm: A Practical Diversification Tool”.  In the paper they introduce two heuristic algorithms for effective portfolio diversification and passive investment management, which they conclude “is an excellent alternative to Risk Parity, Minimum Variance and Maximum Diversification.”

Their algorithm, in my own words, is an alternative asset allocation model which makes intuitive sense. It uses both the historical volatility and correlation of securities in a portfolio to determine asset allocation. Securities with low correlations and volatility relative to the other securities in the portfolio receive higher weightings. The result is a portfolio allocation which changes over time to reflect the evolving volatility and correlations of the securities in the portfolio.

I was happily surprised to discover David freely provides a downloadable “Mincorr spreadsheet” on his site. The spreadsheet uses the algorithms presented in the paper and allows users to calculate their own minimum correlation allocations.

50 Top Stocks

After downloading the spreadsheet from CSS Analytics, I tweaked it to incorporate eight securities (this took a little more time than anticipated). I then created a template to download daily price data to calculate trailing correlations and volatility for the eight securities in the All-Season portfolio (this took a lot more time than anticipated). Finally, I used the Mincorr spreadsheet framework to calculate a daily “minimum correlation” allocation for the eight securities in the All-Season ETF portfolio and I provide the results on the All-Season ETF spreadsheet.

Backtesting this strategy is more difficult than a simple risk-parity or static asset allocation model. The most comprehensive tests are available in the The Minimum Correlation Algorithm paper, which has tests dating to 1980 for a variety of portfolios. The authors of the paper were kind enough to help me run some tests for the All-Season ETF portfolio. The tests are available in the pdf here: minn corr backtests, with tests being run from December 2007 – February 2013. The “min.corr.excel” results used a formula close to the calculation on my All-Season spreadsheet.

The current Minimum Correlation allocations are below:

Name Symbol Minimum Correlation Weightings
Vanguard Total Stock Market VTI 6.71%
 PowerShares DB Commodity Index Tracking Fund DBC 8.86%
SPDR Gold Trust GLD 3.37%
 iShares iBoxx $ High Yield Corporate Bond Fund HYG 29.37%
iShares Emerging Markets USD Bond ETF EMB 9.76%
iShares Barclays TIPS Bond Fund TIP 14.01%
iShares Barclays 20+ Year Treasury Bond ETF TLT 4.71%
iShares Barclays Aggregate Bond Fund AGG 23.21%

Note: my minimum correlation formula differs slightly from the method in the The Minimum Correlation Algorithm in that I use a 20-day historical volatility & correlation versus the 60-day range used in the paper. Using a shorter time period allows the spreadsheet to load quicker, but I may change the 20-day look-back period in the future.

Any errors or omissions in my Min Corr calculation are my own. I have done my best to review and scrub the data, but I make no warranties. As my programming knowledge evolves I hope to post additional minimum correlation backtests.

Past performance is no guarantee of future results. To learn more about the Minimum Correlation algorithm please take a few minutes to visit CSS Analytics.

If you enjoy these free tools, please consider making a donation on the home page of Scott’s Investments using the Paypal link in the upper-right corner!

More on this topic (What's this?) Read more on Barclays at Wikinvest

Gold – Quarterly Update

AlphaClone posted a fourth quarter 2012  summary of institutional activity in Gold (GLD). Quarter over quarter activity shows an overall decrease in institutional holdings of GLD:

However, as Alphaclone notes there are some very prominent funds with positions in GLD:

For those interested in potential applications of institutional holdings see my tests and post from 2010 here.

GLD remains below its 200 day moving average and there is a clear downward sloping channel, as shown by the Finviz chart below, but GLD did find a bid last week at $152:

From a relative strength perspective GLD is trailing two other major asset classes, US Equities and US Bonds (both long-term Treasuries and Aggregate Bonds). I used ETF Replay to test a relative strength system which went long the one ETF with the highest 6 month relative strength among GLD, SPDR S&P 500 (SPY), iShares Barclays 20+ Treasury (TLT) and iShares Barclays Aggregate Bond (AGG), rebalanced monthly. GLD is currently ranked 4th among the 4 ETFs based on 6 month relative strength (TLT is a close third):

For a contrarian perspective, David Banister of The Market Trend Forecast  predicted a cyclical low in Gold for February 2013 in this article, although gold did surpass his low target price for February. A chart from the article below:

 

More on this topic (What's this?)
Mass Exodus from SPDR Gold Fund
Boston Disaster
Read more on SPDR Gold Trust, Gold at Wikinvest

All-Season ETF Portfolio (New!)

I am discontinuing the Basic Portfolio and replacing it with an All-Season ETF Portfolio. First, the Basic Portfolio is being discontinued because there is too much cross-over with the moving average and momentum strategies already used in the Ivy Portfolios and ETFReplay.com Portfolio. I will continue to track the Ivy and ETFReplay.com Portfolios on Scott’s Investments.

The Basic ETF Portfolio will be replaced with an “All-Season ETF Portfolio”. The portfolio draws inspiration from several sources, the first being Ray Dalio, founder of Bridgewater Associates. Dalio created an “All-Weather” investment strategy (Bridgewater pdf available here) with the goal of performing well over all market environments by having exposure to asset classes that perform well in different market environments.

The Fiscal Pop-N-Drop for Equities – Look Out

Below is my proposed portfolio allocation and weighting for an all-season ETF portfolio:

Asset Classes Targeted Allocation Potential Investment Vehicles
Equities 18.75% VTI
Commodities 14.50% DBC, GLD, GSCI Index
Corporate Credit 6.50% VWEHX, HYG
Emerging Market Credit 14.50% PREMX, EMB
Inflation-Linked Bonds 20.75% TIP
Nominal Bonds 25% TLT, IEF

The objective is not to create a one-sized fits all portfolio, but to create a simple portfolio with exposure to different asset classes that perform well in different market environments.

Liberties could be taken with my choice of TLT and IEF to represent nominal bonds. A more broad-based bond ETF like AGG, BND or BOND could serve as a viable substitute. Corporate credit could also be divided among high-yield (HYG) and investment-grade corporate bonds, using  an ETF like LQD. Commodities are a broad sector, Gold may be the most popular among the investment masses but several other ETF options exist for broader commodity exposure.  Equity exposure could be further allocated among market-capitalization, sectors, global exposure, value vs. growth, etc. Inflation-linked bonds could be further divided among duration as well as global exposure through an ETF like WIP. The possibilities are nearly infinite, but the objective was to keep it simple.

I took the portfolio above and tested it using ETFReplay.com. The nominal bonds were equally split between TLT and IEF (12.5% each) and the commodity exposure was split equally between DBC and GLD. In cases where the ETFs listed above had trading histories beginning later than 2005, I substituted the listed ETF or Index for testing purposes.

The tests ran three variations of the proposed all-season portfolio. The first was a buy-and-hold with an annual rebalance. The second used a 10 month simple moving average rule to sell any of the funds or ETFs when they crossed below their 10 month moving average and purchased them when they crossed back above it (re-balanced monthly). The third used the same 10 month SMA rule but used an equal weighted version of the portfolio. Comparisons to SPY, AGG, VBINX, and a variation of the Ivy 5 Portfolio, which uses the same 10 month SMA crossover system, are presented:

2005-2012
Strategy CAGR Total Return Volatility Sharpe Max Drawdown Trades
All-Season ETF Buy and Hold, Annual Rebalance 8.10% 85.80% 6.10% 0.83 -19.67% annual
All-Season ETF (Mutual Funds/ETFs) w/ 10 Mo SMA 7.40% 76.30% 4.90% 0.96 -4.60% 128
All-Season ETF (Mutual Funds/ETFs) Equal Weighted w/10 Mo SMA 7.70% 80.80% 5.40% 0.91 -6.50% 132
VBINX (Vanguard 60/40 Balanced Fund) 5.30% 51.10% 13.00% 0.23 -36% N/A
SPY 4.20% 38.60% 22.10% 0.16 -55.20% N/A
AGG 5.20% 49.60% 5.60% 0.45 -12.83% N/A
Ivy 5 w/ 10  month SMA (AGG, EFA, SPY, VNQ, GSCI Index) 6.10% 61.10% 9.90% 0.35 -14.70% 81

For the 2010-2012 test only ETFs were used:

2010-2012
Strategy CAGR Total Return Volatility Sharpe Max Drawdown Trades
All-Season ETF Buy and Hold, Annual Rebalance 10.80% 35.80% 5% 1.84 -3.28% annual
All-Season ETF (ETFs only) w/ 10 Mo SMA 6.70% 21.40% 5% 1.06 -3.70% 56
All-Season ETF (ETFs only) Equal Weighted w/10 Mo SMA 6.20% 19.60% 5.40% 0.89 -4.20% 56
VBINX 9.20% 30.20% 10.80% 0.73 -10.87%
SPY 10.80% 36.00% 18.40% 0.57 -18.61%
Ivy 5 w/ 10  month SMA (AGG, EFA, SPY, VNQ, GSCI Index) 2.10% 6.30% 11.70% 0.12 -14.70% 45

The portfolio has been a low volatility alternative to both a balanced mutual fund (VBINX), the Ivy 5 Portfolio, and equities (SPY). The SMA variation has under-performed relative to equities since 2010, although volatility and drawdowns remained low relative to the other benchmarks.  Since 2005 the portfolio has shown significantly lower drawdowns and volatility then any of the benchmarks. Not shown in the tables above is the buy-and-hold All-Weather Portfolio has a correlation of .39 to AGG from 2005-2012 and .57 to VBINX.

Past performance is no guarantee of future results. Factors that would impact the test results above are commissions and taxes. The more transactions, the higher the potential commissions and taxes.  These factors, among others, should be considered when evaluating the buy-and-hold approach to the 10 month SMA approach.

Readers may be concerned about exposure to long-term government bonds. It is a fair concern in the present monetary and fiscal environment; however, the objective of the portfolio is to provide exposure to a variety of assets that perform differently in different market environments. Its purpose is not to predict when these changes in market conditions will occur. In this aspect, it is not dissimilar from the Permanent Portfolio, which I have profiled on several occasions. AGG, BND, or PIMCO’s BOND offer some additional diversification within the bond sector.

In the next few days I will post the portfolio as a spreadsheet so that readers can follow the performance at Scott’s Investments. The spreadsheet will track portfolio allocations and 10 month SMA signals. As a bonus, I will list some potential commission-free ETFs (depending on your broker and a variety of other factors) to substitute for the ETFs listed above.

For some further research on the All-Weather portfolio CSS Analytics posted a worthwhile piece in November.

Ivy Portfolio for October

Early in 2012  Scott’s Investments added a daily Ivy Portfolio spreadsheet. This tool uses Google Documents and Yahoo Finance to track the 10 month moving average signals for two of the portfolios listed in Mebane Faber’s book The Ivy Portfolio: How to Invest Like the Top Endowments and Avoid Bear Markets. Faber discusses 5, 10, and 20 security portfolios that have trading signals based on long-term moving averages.

The Ivy Portfolio spreadsheet tracks the 5 and 10 ETF Portfolios listed in Faber’s book. When a security is trading below its 10 month simple moving average, the position is listed as “Cash”. When the security is trading above its 10 month simple moving average the positions is listed as “Invested”.

The spreadsheet’s signals update once daily (typically in the evening) using dividend/split adjusted closing price from Yahoo Finance. The 10 month simple moving average is based on the most recent 10 months including the current month’s most recent daily closing price.  Even though the signals update daily, it is not an endorsement to check signals daily. It simply gives the spreadsheet more versatility for user’s to check at their leisure.

Now you can follow me on Stocktwits and Twitter!

The page also displays the percentage each ETF within the Ivy 10 and Ivy 5 Portfolio is above or below the current 10 month simple moving average, using both adjusted and unadjusted data.

If an ETF has paid a dividend or split within the past 10 months, then when comparing the adjusted/unadjusted data you will see differences in the percent an ETF is above/below the 10 month SMA. This could also potentially impact whether an ETF is above or below its 10 month SMA. Regardless of whether you prefer the adjusted or unadjusted data, it is important to remain consistent in your approach.

I do not track these portfolios as hypothetical portfolios like I do with other portfolios on the site. However, I will periodically post backtest results on the strategy. Below are updated backtest results for the Ivy Portfolio using ETFReplay.com.

The backtest results for the Ivy 5 Portfolio since 2007 and 10 month simple moving average with a monthly update are charted below. For the backtests, iShares Barclays Aggregate Bond (AGG) was used in lieu of BND and iShares MSCI EAFE (EFA) was used in lieu of VEU because they have longer trading histories:

The Ivy 10 Portfolio, using a 10 month moving average and updated monthly has performed as follows since 2007 and compared to SPY. Again, AGG and EFA were used in the backtests:

The strategy’s strength is avoiding significant drawdowns during periods of market turbulence, such as 2008. During periods of strong uptrending equity markets it has the potential to under-perform a benchmark such as SPY.  ”Choppy” markets, in which markets are trend-less can also reduce the strategy’s returns as securities bounce above and below long-term moving averages without establishing a trend.

The current signals based on September 28th’s closing prices are below.  Real-estate linked ETFs and US Equity ETFs remain the strongest sector in terms of their percent above their 10 month moving average. All of the securities in the 5 and 10 ETF portfolios are above their 10 month moving averages.

The first table is based on adjusted historical data and the second table is based on unadjusted price data:

Backtests Galore (or, beating Mr. Market)

A reader asked if I had updated backtest results for a variety of ETF Replay and Ivy Portfolio strategies.  I track one tactical ETF asset allocation strategy live on Scott’s Investments using data from ETFReplay.com; however, from time to time I will backtest the results and also show how adding different re-balancing rules impact the results.

The Ivy Portfolio strategy’s long/cash signals (background here) are also tracked on Scott’s Investments, but the portfolio is not tracked live. However, it is easy enough to backtest the results using ETFReplay.com.

Now you can follow me on Stocktwits and Twitter!

Below are backtest results for a relative strength ETF strategy. Returns include dividends but exclude commissions, taxes, and slippage (in other words, real results will differ). The buy/sell strategy for the portfolio is simple: purchase the top  ETFs based on a combination of their 6 month returns, 3 month returns, and 3 month volatility (lower volatility receives a higher ranking) and the average of  the 3 month return, 20 day return, and 20 day volatility.  I refer to these two different sets as “6/3/3″ and “3/20/20″.  The top 2 ETFs in the 6/3/3 ranking and top 2 in the 3/20/20 ranking are purchased each month.

The backtest below is not identical to the ETFReplay strategy I track on a monthly basis. The difference is that the backtest does not always hold 4 ETFs – if the top 2 ETFs in the 6/3/3 and 3/20/20 rankings are the same, then the backtest holds only 2 (or 3) ETFs.   However, the two strategies are similar and there are several months when both the backtested strategy and the one I track will hold identical positions.

It is also important to note that not all 25 ETFs were available at the beginning of 2005. Less than half were available at the start of the test and not until the start of 2009 were all 25 available.

The first backtest are based on a monthly rebalance beginning in 2005:

The second test is based on a semi-monthly rebalance:

While semi-monthly re-balancing appears to bolster returns, keep in mind that returns exclude commission costs and taxes. More frequent re-balancing will tend to increase the costs associated with both.

The next set of tests refer to the Ivy Portfolio strategy detailed in Mebane Faber’s book The Ivy Portfolio: How to Invest Like the Top Endowments and Avoid Bear Markets. I also maintain a spreadsheet which tracks these signals daily and also typically write a monthly review (this month’s review is available here).

When a security is trading below its 10 month simple moving average, the position is sold and cash is held until the next month. When the security is trading above its 10 month simple moving average cash is used to invest in the security.  For the purposes of these tests I used a 5 ETF portfolio consisting of AGG, DBC, VNQ, VEU, and VTI and 10 ETF portfolio consisting of AGG, DBC, GSG, RWX, VNQ, TIP, VWO, VEU, VB, and VTI.

(note: For the purposes of this test I substituted AGG for BND due to the fact that AGG had a longer trading history. BND is tracked on my Ivy Portfolio spreadsheet and is closely correlated to AGG).

Below are the results of a monthly re-balance of the 5 ETF portfolio using a 10 month simple moving average system. This test was run from 2007-present with 2007 being selected as the starting date due to DBC’s shorter trade history. All five ETFs were available at the start of this test:

The next test uses the same rules but tests a portfolio of 10 ETFs. Only 7 of the 10 ETFs were available at the start of 2007, making comparisons to the 5 ETF portfolio incomplete:

The Ivy Portfolio system did a good job of avoiding the significant drawdowns of the 2008. In theory the system may miss the early moves of a bull market, potentially limiting total returns in a prolonged bull market but has historically limited drawdowns and volatility.  It could also suffer from whiplash when a position hovers above and below its 10 month moving average for an extended period of time.  Commissions and taxes will also further reduce returns.