Jarrow and Kwok, in their new paper "Riding A Bubble: A Study of Market-timing Trading Strategies," identify when there are Q-bubbles based on local martingale properties. The idea is that a bubble will exhibit extreme values over different time horizons, and an investor should hold a significant bubble move until it reaches a set barrier. At this time, the investor should exit. Ride the bubble until the returns reach an extreme and then walk away. The basic story seems easy enough, yet the key is to determine the bubble component, which is based on the tail probabilities. Bubbles have a fat-tailed Pareto distribution. If an investor sets an upper bound on the price of the asset and it is reached before a certain time, then exit.
The idea is relatively simple, yet despite the simulations run in the paper, this point of exit is harder to find in practice. It will encourage getting out of positions early, even if there is an optimization and an accounting for risk aversion.
The trend-follower will generally not follow this type of strategy. The trend-follower will always hold the position until there is a reversal and a stop is hit. You will sacrifice some of the return in exchange for carrying any position as long as possible. Yes, there will be losses in the end when the market turns, but the ability to maintain a position in a bubble will generally be worth the added risk and the likelihood of some give-back.
No comments:
Post a Comment