Selling rules work. History shows that a policy of selling whenever prices fall below their 10-month (or 200-day) moving average would have increased long-run average returns on UK, US, Japanese and emerging market equities. The rule does not, however, work even over the long-run for all assets. In particular, it has been no use for gold investors and of limited use when applied to the Nasdaq composite.
To test this, I asked how it would have worked for both, had we applied it at the end of every month since December 1985, assuming that you would have made zero returns in those periods when you were out of either asset. The answer is: badly for gold and indifferently for the Nasdaq.
A buy-and-hold investor in gold would have made just under 5 per cent in US dollar terms, and just over 5 per cent in sterling terms. In both currencies, an investor following the 10-month rule would have done worse. Yes, such an investor would have seen lower volatility. But even so, volatility-adjusted returns (as measured by the Sharpe ratio) would have been lower.
The one merit of the 10-month rule is that it would have saved us from the worst losses. In 2013, a buy-and-hold investor would have lost over 27 per cent in dollar terms. But someone using the buy-and-hold rule would have lost just 17.4 per cent in US dollars and nothing in sterling terms.
| Performance of the 10-month average rule | ||||
| Gold (£ terms) | Nasdaq composite | |||
| Buy & hold | 10-month rule | Buy & hold | 10-month rule | |
| Return | 5.2 | 3.3 | 11.5 | 9.4 |
| Volatility | 16.8 | 13.2 | 23.9 | 18.7 |
| Sharpe ratio | 0.31 | 0.25 | 0.48 | 0.50 |
| Worst loss | -29.1 | -15.6 | -59.8 | -40 |
| Based on monthly returns since December 1986 | ||||
Nor would the rule have done spectacularly well when applied to the Nasdaq composite. It would have made you two percentage points a year less than a buy-and-hold strategy. That doesn’t sound much, but it compounds a lot over 36 years. Returns would however have been less volatile, meaning that the Sharpe ratio would have been a touch better.
The rule would, however, have saved us from some savage losses. In the year to September 2001, for example, the Nasdaq lost almost 60 per cent. But the 10-month rule would have got us out of the market beforehand, saving us this loss. It would also have got us out before the near 40 per cent losses in 2008-09. Overall, though, these huge savings come at a price: the rule would have got us out of the Nasdaq on many times before it bounced back, causing us to miss some nice profits.
There are reasons why the rule has been so lacklustre in both assets.
In the case of gold, it’s because we’ve not seen the sort of long and deep bear markets that equities suffered in 2000-03 and 2008-09. It is on these occasions that the rule works well.
In the case of the Nasdaq, we have seen such bear markets and the rule did well then. It’s just that they have been rare exceptions to a long-term uptrend, during which buying on dips would have worked.
All this reminds us of the limitations of the rule. It works well when markets see momentum-driven selling, so that cheap asserts get even cheaper. But it fails when investors buy on dips.
For most of the time, buy-on-dips investors have predominated in both assets, and so the rule has done badly.
But, but, but. As MIT’s Andrew Lo has pointed out, markets are like ecosystems whose populations change over time. The proportions of “buy on dip” investors and “sell on momentum” investors vary over time.
My chart shows this for the case of gold, showing the relative performance of buy-and-hold versus the 10-month rule. From 1985 to the mid-1990s the two performed roughly equally well, suggesting that buy-on-dippers and momentum investors were roughly equally balanced. In the late 1990s, though, the 10-month rule out-performed suggesting the momentum investors dominated. And then from around 1999 to 2012 the 10-month rule underperformed, implying that the buy-on-dippers were in the majority. Since then, the two strategies have done roughly as well as one another, implying rough equality again between buy-on-dippers and momentum investors.
Herein lies a case for using the 10-month rule. It is possible that momentum sellers will one day get the upper hand again. When they do, the 10-month rule will pay off.
In fact, there’s another case for the rule. Even in these two cases where it hasn’t generated great returns it has protected us from big losses. If you’re an especially loss-averse investor whose main priority is the preservation of capital, there is therefore a case for you to follow it.
In the case of gold, though, I’m not sure we should do so. We shouldn’t regard it as an asset to trade. Instead, its virtue lies in its ability to diversify risk. During times of crisis investors seek safe havens and dump sterling. In such circumstances, gold does well. It is therefore a form of insurance against recessions and big losses on equities. And the point about insurance policies is that it is a good idea to always have them rather than try to predict when you’ll need them and when not.
My story here, though, is not just about gold and the Nasdaq. It’s about the nature of financial markets. The ecology of them varies from market to market and from time to time: the proportions of buy on dip investors and sell on loss of momentum investors varies. Because of this, rules that work well in some times and places don’t work in others. Economics is not like physics: there are few robust general laws.
