Please Note: Blog posts are not selected, edited or screened by Seeking Alpha editors.

Creating Alpha Using Dynamic Weight Allocation

Summary

Many portfolio managers are rebalancing their portfolios using the equal weighting scheme. We briefly mention the pros and cons of this approach.

We dynamically adjust the weights of our Model Portfolio Strategy with nine different factors and compile the return statistics for each of them.

We found tilting weights of our positions by lowest Enterprise Value (Market Cap + Total Debt - Cash Equivalents) creates value added.

All the Model Portfolio Strategies we created so far are equal weighted. We are holding 25 stocks and each of them have an initial weight of 4%. We then let the market determine future weights with price fluctuations. The only weight constraint we use is to trim stocks when weighting exceeds 10%.

With this approach, winners become larger and losers smaller, therefore creating a momentum effect in the portfolio. Drawbacks of this method are twofold: 1) portfolio returns may become significantly distorted by the winners and 2) the portfolio is too reliant on these winners to generate future return.

What if we tilted weights intelligently using various factors and rebalanced our positions accordingly? In this piece, we will test dynamic weight allocation, a relatively new quantitative portfolio strategy. We will determine if this technique improves returns over the traditional equal weighting scheme.

You will find, in the table below, our initial equal weighting serving as the benchmark to beat and nine factors we hypothesized as being useful to adjust the weights of our positions. For each factor, we provide a short definition of how we quantified the factor and the rationale behind their use.

Let’s define the parameters used in each simulation. First, stocks are all US large caps chosen from the S&P 500 universe. Second, we backtest the 1999-2017 period to cover two business cycles and use all data available. Third, we set the position weighting constraints at 2.5% min and 10.0% max.

The rebalancing period and reconstitution period will be quarterly for each backtest. Rebalancing means we are going to adjust weights of each position based on the chosen factor. Reconstitution means we are going to check if a security needs to be replaced. Here is the statistical summary of our findings:

In conclusion, four factors used for dynamic weight allocation generates a Sharpe Ratio above the equal weighting scheme. Two of them are worth looking at: Enterprise Value generates 110bps alpha over equal weight with a win rate of 84.2% and Correlation 50bps with a maximum DD of -36.1%.

In all cases, the portfolio turnover increased by at least 14.3% by dynamically adjusting weights. Is this increase of portfolio turnover worth the hassle? We believe so for the Enterprise Value weighting scheme but maybe not for the Correlation version as it created the highest portfolio turnover at 88.3%.

For those interested to learn more about our US Model Portfolio Strategy discussed in this monthly research, you can access two-page fact sheets in the Strategies section of our website. If you register for a free membership, you will have access to thirty-page presentations in the same section.

Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.