In
Part I of this series the frame work for a basic Tactical Asset Allocation strategy was discussed using two ETF's representing stocks (first SPY, later MDY) and bonds (TLT). Following on the daily signals of the first post, the continuation of this series will be all about the higher time frames, starting with weekly signals.
these days. And as it happens, the AFL-code provided in the previous post is just as suitable for weekend traders too.
Weekly Periodicity
As a starting point, the weekly equivalent of 85 days for the momentum calculation is used: 17 weeks without any smoothing.
 |
| Portfolio Equity SPY - TLT (17 weeks), draw down periods in red |
 |
| Profit table SPY - TLT (17 weeks) |
| SPY-TLT (17w) | | 2004 - 2013 |
| Total Profit | | 211 | % |
| CAR | | 12.05 | % |
| MDD (trades) | | -16.16 | % |
| MDD (system) | | -16.16 | % |
| Calmar (0%) | | 0.75 | |
| Sharpe (0%) | | 0.76 |
|
| Trades | | 43 | |
| Winners | | 58.14 | % |
Optimizing for x-weeks sorted on the Calmar ratio (CAR/MDD) AmiBroker renders the below overview.
 |
| Performance of SPY - TLT on x-week momentum |
Applying a y-week simple moving average to smooth the price data next to the x-week momentum period, AmiBroker's optimizer finds (x,y) combinations with better CAR, fewer trades and a very high Win% (up to nearly 90%, see highlighted rows in blue).
 |
| Performance of SPY - TLT on x-week momentum with y-week price smoothing |
 |
| Calmar distribution for SPY - TLT on x-week momentum with y-week price smoothing |
|
What would happen if SPY was replaced by MDY, like was done in the previous post? Better Calmar, better CAR, same MDD and high Win%.
 |
| Performance of MDY - TLT on x-week momentum with y-week price smoothing |
 |
| Calmar distribution for MDY -TLT on x-week momentum with y-week price smoothing |
Universe expansion
One of the commenter's on Seeking Alpha,
A.L., suggested to expand the strategy's two-asset-universe with EEM, the iShares Emerging Markets ETF. While A.L.'s vehicles thus are SPY, EEM and TLT, in this backtest MDY, EEM and TLT are used instead. Besides the SPY-MDY swap, price data smoothing is applied too.
 |
Performance of MDY -EEM - TLT on x-week momentum with y-week price smoothing sorted on Win%
(max CAR% highlighted in blue) |
 |
| CAR% distribution of MDY -EEM - TLT on x-week momentum with y-week price smoothing |
MDY - IEV - EEM - DBC - TLT
To conclude the discussion of the weekly time frame, let's take the universe expansion one step further by adding ETF's for European stocks and commodities as well: IEV and DBC (
back-extended with index data). The mountain graph for this five asset class universe is the below, with detailed stats for the highest Calmar ratio combination.
 |
| CAR% distribution for five asset class universe on x-week momentum with y-week price smoothing |
 |
| Portfolio Equity five asset class universe (11w/9w), draw down periods in red |
 |
| Profit Table five asset class universe (11w/9w) |
| N=5 universe (17w/9w) | | 2004 - 2013 |
| Total Profit | | 1213 | % |
| CAR | | 29.40 | % |
| MDD (trades) | | -20.11 | % |
| MDD (system) | | -20.11 | % |
| Calmar (0%) | | 1.46 | |
| Sharpe (0%) | | 1.15 |
|
| Trades | | 46 | |
| Winners | | 71.74 | % |
To be continued in part III
In the final part of this series portfolio performance is examined for the monthly and quarterly time frames. So stay tuned and in the meantime please share your thoughts and/or idea's in the comment section. Special thanks goes to the eight generous supporters for their donations. Any contribution is very much welcomed.
