Trend Following

Trend following commodity strategies attempt to derive future expected performance from past historical performance. This involves two steps. We first need to identify whether a trend has been established and secondly for how long it will continue.

In essence, a trade follower will always be late to hop on a trend and will almost invariably be surprised when the trend ends (often as the result of an exogenous, i.e., unforecastable shock). Only if the trend continues for long enough to cover the costs from a trend reversal, will the strategy be profitable.

Return momentum strategies are the most common strategies in the commodity universe. They come in the form of either simple momentum (buy winners and sell losers) or crossover momentum (buy a commodity if the short run performance exceeds the long run performance).

Generically we can express a trend following strategy as mom (h, s, l), where h denotes the holding period horizon, s the short-run moving average, and l the long-run moving average.

For example, mom (3, 6, 12) denotes a momentum strategy that will invest into a commodity, if the 6 months moving average exceeds the 12 months moving average and vice versa. As the strategy is good for the next 3 months, the question arises, "what do we do after 1 month?" After all, a new signal arrived.

Do we want to create an entirely new portfolio, throwing out our old 3 month view? Given that our holding period assumption is 3, we will for each period build a portfolio that is a mixture of the past three momentum portfolios.

Equal weighting stacked portfolios implicitly assumes that there is linear decay in information ratio. While this is not necessarily true, it seems to be a robust assumption.

What are the economic foundations of trend following strategies? We start with behavioral finance models. Suppose we have a market with two types of investors, both exhibiting bounded rationality.

One type of investors only reacts to fundamental information, disregarding the information in price changes, while the second type of investors disregards fundamental information and only reacts to price changes. If fundamental information spreads slowly, we will see commodity prices to initially underreact to the arrival of new information.

This will kick-start momentum traders that have observed past prices to rise. Sitting on a self-accelerating strategy, momentum traders continue buying in an attempt to arbitrage the slower fundamental investors. Effectively this will lead to a market that shows both initial underreaction and final overreaction.

A second explanation for the success of trend following commodity strategies is their link to business cycles. Given that commodities are closer to consumption goods than to assets, they are unlikely to be priced by a forward-looking discounting mechanism.

As such they should be much more sensitive to changes in business cycle conditions as they lack the ability to look through to the future. Finally, it is intuitive that momentum strategies work best where it is hard for fundamental models to find fair values and as such learning from past prices is more widespread.

Trend Following
Trend Following