Rebalancing

One piece of arithmetic used in two settings. A holder of several positions rebalances when relative performance has pulled the mix away from the weights it was built to, since the best performing holding grows into a larger share of the total without anything being bought. An index provider rebalances on a published schedule, resetting constituent weights to whatever its weighting rule specifies, and separately reconstitutes by adding and removing constituents. A fund tracking that index has to trade to match, which is why volumes cluster on rebalancing dates.

The trigger is either a calendar or a band. A calendar rule acts at fixed intervals whatever the drift; a threshold rule acts only once a weight has moved a stated distance from its target; many arrangements combine the two by checking on a calendar and acting only if the band is breached. The trade itself is mechanical once the rule is set, which is the point of the rule: the weights, not a judgement about the holdings, decide what is bought and sold.

Two things are contested and one is not. It is not contested that rebalancing has costs, since every adjustment pays a spread and any commission, and that predictable, published index rebalancing dates are traded around by other participants. What is contested is the so called rebalancing bonus: the claim that the discipline adds return rather than only controlling drift holds under particular assumptions about markets reverting, and the academic literature does not agree that those assumptions generally hold. Products carrying a daily reset, such as a leveraged exchange traded fund, are a separate case entirely: their reset is a rebalancing that occurs every day, which makes their result over a longer period depend on the path prices took and not only on where they finished. The loss over such a period can be larger than the stated multiple of the underlying index's own decline, and the gain can be smaller than the same multiple of its rise.