This post is available as a PDF here.
- We break portfolio construction into two unique phases: signal generation and the rules that determine allocations.
- We use the analogy that this process is like cooking, where our signals are the ingredients and our allocation rules are the recipe.
- While most firms focus their research on generating the best ingredients, we believe it is important to acknowledge up front that the ingredients will spoil from time-to-time
- We believe that thoughtful recipe design can help reduce the portfolio impact from spoiled ingredients.
At Newfound, we talk about portfolio construction as having two distinct, but equally important pieces.
To explain the two pieces, we use a hokey analogy: portfolio construction is a lot like cooking where there are ingredients and there are recipes.
You can have the best ingredients in the world, but a bad recipe will ruin their value.
Similarly, it does not matter how good your recipe is if all your ingredients have spoiled.
In portfolio construction, we say that the signals or metrics you generate about an investment are the raw ingredients. For example, at Newfound we use models that seek to determine if an asset class is exhibiting positive or negative momentum.
Without a recipe – the rules for combining ingredients – the ingredients themselves are fairly useless. We need to take our positive and negative momentum signals and turn them into allocations.
This week we wanted to share an example of just how important the choice in recipe can be.
In our “recipe,” we choose to dollar cost average our tactical trades over a 4-week period. Our research shows that this approach helps mitigate the impact of whipsaw without sacrificing significant portfolio adaptability.
This year, the U.S. large-cap utilities sector has exhibited significant mean-reversion, with several 1000+ basis point moves. This is exactly the type of environment where momentum signals “spoil,” signaling to buy at highs and sell at lows.
Applying our momentum signal on a weekly basis, we can see that whipsaw would have caused in excess of 400 basis points of under-performance.
However, using our dollar cost averaging method – relying on the recipe to help account for spoiled ingredients – we significantly reduced this loss:
While the signal was wrong, the recipe to dollar cost average completely eliminated the whipsaw cost.
We think the utilities sector in 2015 highlights a perfect example of an environment where a momentum model is expected to struggle.
One approach to managing this risk is not to add more bells and whistles to our momentum model in an effort to improve its accuracy. In our analogy, this is akin to trying the impossible task of trying to create ingredients that never spoil. While we may be able to improve our accuracy, the expectation of 100% accuracy is unrealistic. Operating on the assumption that we have 100% accuracy is downright dangerous.
So instead we acknowledge up front that ingredients will spoil from time to time and try to use a thoughtful recipe design – in this case, dollar cost averaging – as a means to manage this risk.