Time Series Momentum


Over the past few days I have been writing about strategies that buy the currencies that are going up the most strongly and sell the currencies that are going down the most strongly.

I showed how some simple strategies based on this principle would have performed over approximately the last 21 years.

For example, the 3-month version of this strategy yielded 94.45% over a 21.75-year period (meaning each trade made a profit of 0.36% on average). The maximum peak to trough draw-down was approximately 27%.

Here is another variation: instead of picking the strongest and weakest currencies to trade, how about trading all of the currencies and just buying each currency when it is going up and selling each currency when it is going down?

This kind of strategy is known as “time series momentum”, as opposed to “best of momentum”. It just means that the price has to be higher or lower than it was, not outperforming or under performing any other currency.

Applying a 3 month look back period here over the same overall time period, the strategy returned 357.24% (meaning each trade made 0.20%). The maximum peak to trough draw-down was approximately 131%. The equity curve is shown below:

The maximum peak to trough drawdown was approximately 131%. The maximum peak to trough draw-down was approximately 131%.

As you can see, the time-series strategy had a larger proportionate draw-down (36%) than the best-of strategy (27%). It also made a higher average return per trade (0.36% compared to 0.20%). So the best-of momentum strategy is better, right? No, not necessarily.

The reason for this is that best-of strategies built from time periods other than 3 months, do not perform as well as time-series strategies built from other time periods. This suggests that time-series strategies are more robust i.e. more likely to keep working. This is probably because with best-of strategies you are putting all your eggs in one basket, whereas time-series strategies will keep you more diversified.

To give an example, the 6 month look back applied to a time series strategy over the same time period gave a very similar return to a 3 month look back. However, a 6 month look back to a “best of” strategy actually produced a loss overall of -56.02%.

Adam is a Forex trader who has worked within financial markets for over 12 years, including 6 years with Merrill Lynch. He is certified in Fund Management and Investment Management by the U.K. Chartered Institute for Securities & Investment.