It’s a public holiday today in the world’s two major Forex trading centers (London and New York), so it is not a great surprise that today’s market movements are thin and light. Monday is typically a relatively quiet day in the market, anyway. As there is not a lot going on, I thought I would try to expand a little more on what I wrote yesterday about trend following, or “trend trading” as it is often called.
I already outlined the major pros and cons of trend trading for retail traders, then went on to present a long-term back test of a strategy where long or short trades in a basket of assets are entered depending upon whether the price of an asset has gone up or down over the previous month. The strategy would have produced an annualized return of 11.78%, but with a very large maximum draw-down of 120%, which makes such a strategy difficult to trade. There are a couple of interesting places this observation can lead us to.
Firstly, maybe 1 month is an overly short-term period. Couldn’t we do much better if we traded with the long-term trend? To find out, I applied a filter where trades are only taken if the price has moved in the same direction over the previous 12 months as it has over the previous 1 month. Applying this filter, the return from 2001 to 2016 decreases from 443.60% to 365.71%, which works out to an annualized return of approximately 10.65% instead of 11.78%. This number is better than it looks, for this strategy is in the market less than the 1 month strategy, so the costs of spreads, commissions, and overnight financing are all lower. The worst draw-down, was only 93% compared to 120%. The total return for every asset in the basket, including each Forex currency pair, was positive. It should be noted however, that the Forex assets produced the lowest returns: further proof of the relatively low propensity to trend of fiat currencies. If we remove the Forex assets, we get a better risk-adjusted return: 316%, with a maximum draw-down of only 54%. I am not the first author to note that while diversification tends to lower risk in “normal” markets, during periods of great shock such as 2008, it seems that all the trends can unwind at once, producing surprisingly large losses. The chart below shows all assets in blue, and non-Forex assets in Gold:
Another issue which I will get into later is how results can usually be improved by adjusting position sizing to the current volatility of each asset. This works because volatility tends to cluster.