Source: http://www.advisorperspectives.com/dsho ... Part-2.php
The 25/25/25/25 PP makes sense both from a back-testing perspective and a theoretical perspective, which is about as much as we can ask for from a portfolio strategy. It's a great starting point. Let's see if it can be improved with methods that align well with HB's philosophy.
Adjustment #1: Using a Risk Parity approach to weigh the asset allocation not by capital, but by the volatility of each asset class.
Alignment with "Don't try to time the market": 10/10. There is no connection between Risk Parity and trying to predict future events.
Alignment with "Be prepared for all economic conditions": 10/10. This approach takes the exact same 4 asset classes that HB proposed, but instead of simply using an equal capital weighting, it adjusts that weighting based on the volatility of the four asset classes.
Apparently this approach improves the Sharpe ratio in almost any approach to investing, meaning volatility will decrease at a similar target return or target return will increase at a similar target volatility. The link shows that this also works for the PP when back-testing.
How I need your help:I'd like to test this on more than a single period (1970-2012) to see if it is "always better" for each decade on its own, for example. However, I am having trouble determining exactly what the process for allocating and rebalancing would be. If anyone has any guidance on how to do this, I would much appreciate it so I can run the tests in a more robust way than the link did. One approach would be to simply allocate based on historical volatility of each asset - e.g., if the volatilities were 10%/5%/5%/2%, the allocation would be 10%/20%/20%/50%. Another is to only look at down-side volatility instead of overall volatility (i.e., more of a Sortino than Sharpe approach), but I haven't thought through the mathematical implications of this. Lastly, there is the question of whether or not volatility should be static or adjusted on a periodic basis as we move into the future.
Adjustment #2: Using Faber's 10-month simple moving average (SMA) momentum approach to improve returns at a given level of risk.
Alignment with "Don't try to time the market": 5/10. While this approach is definitely "timing the market", it does not do so based on hunches but rather using a systematic approach that is equally applied to all asset classes in the mix. Less of a prediction approach than a systematic approach that has done well in back-testing and makes some conceptual sense.
Alignment with "Be prepared for all economic conditions": 8/10. This approach takes the exact same 4 asset classes that HB proposed, but instead of simply using an equal capital weighting, it adjusts whether or not one is invested in a specific asset class or cash based on a 10-month price trend. The risk is missing out on the first several months of a specific economic condition (e.g., boom, deflation, etc.), but if HB is right in saying that the asset class that wins during any specific economic conditions "wins big", then the lag should not be substantial and the avoided losses should outweight the foregone gains.
This approach departs from HB's approach of always being invested in all four asset classes, but back-testing shows that this can improve the risk-return characteristics of the PP (see link) when also combined with Adjustment #1.
How I need your help:Trying to figure out exactly what the process would be for allocating and rebalancing when combining Adjustments #1 and #2. If I can create a clear logic to how that would look, I can back-test across various scenarios instead of just the 1970-2012 time period.
In conclusion, I know that these ideas may not be popular with the majority of the forum but I think the link is worth a read and the ideas worth some exploration.
Enhancing the PP by adding Risk Parity and/or Faber's Timing Model
Moderator: Global Moderator
Re: Enhancing the PP by adding Risk Parity and/or Faber's Timing Model
I'll take a shot. First, I'd like to see something more in the 18-24 month range in addition to shorter periods when checking for momentum. This way is more likely riding major macro trends. The momentum rebalance point would also be a good time to do volatility recalibrations at maybe something like a 1:2 or 1:3 ratio of momentum look back to volatility look back.
I'm advocating longer look backs simply because shorter periods haven't really helped much. With this particular mix of assets I've found the volatility adjustments to usually be exactly wrong at the exactly wrong time. Volatility goes down/port weight goes up on bullish assets and the reverse on bearish...then the inevitable turn happens and for forward performance you are 180 from what you should be...thus there is little improvement on the standard 4x25 model. It's really the same as hitting a bottom or top in stocks and literally everything swaps places in terms of forward performance (momentum crush).
I'm advocating longer look backs simply because shorter periods haven't really helped much. With this particular mix of assets I've found the volatility adjustments to usually be exactly wrong at the exactly wrong time. Volatility goes down/port weight goes up on bullish assets and the reverse on bearish...then the inevitable turn happens and for forward performance you are 180 from what you should be...thus there is little improvement on the standard 4x25 model. It's really the same as hitting a bottom or top in stocks and literally everything swaps places in terms of forward performance (momentum crush).