Rebalancing (Part 1)

Rebalancing is one of those things that always seemed to make intuitive sense to me although I've only recently developed a more in-depth appreciation for the consequences upon one's portfolio. Basically, rebalancing just means selling a portion of stocks that have done well and using the proceeds to buy more of the stocks that have lagged behind, in order to make the weighting of each stock in the portfolio more in line with your targets. If your portfolio is constructed to simply hold the same stocks as an index, then there is no rebalancing required because both the index weights and the weightings in your portfolio will drift together over time as certain stocks outperform others. However, if your target weights are fixed (e.g. equal weighting for each stock), then you will need to periodically rebalance lest your actual weightings diverge so much from the target weights that you become uncomfortable with the portfolio composition.

So what are those consequences of rebalancing? To answer that, it's helpful to create a model of what would have happened over the past 10 years if you had a 10-stock portfolio and rebalanced it every week back to equal weightings vs. if you had not rebalanced and simply let the weightings drift (i.e. Buy&Hold strategy). The three charts at right and bottom provide the results of this test. The stocks used in the model were taken from the list below of the largest companies in each economic sector.

Next week I'll discuss the results of this model and see if there are any insights to be had about rebalancing. We may also take a look at another model of the same thing, except rather than using the price histories of 10 real stocks, we'll construct a synthetic portfolio using the random number generator function in Excel, which will allow us to easily sensitize the results for varying levels of the artificial stocks' volatility, correlation, and returns. This way, we'll be able to see if rebalancing 'should' be beneficial in theory or if the model using real stock data is just a fluke based on the idiosyncrasies of those particular stocks.