If shares can rise too much they can also fall too much, and for the same reasons: herding and momentum.
But how can we tell what is 'too much'? Stock market cliches warn us to be wary of dead cat bounces and of trying to catch falling knives. And the cliches are right: in volatile day-to-day price moves, it's hard to distinguish between noise and signal.
However, there's a simple rule that can help us here. It has been proposed by Mebane Faber of Cambria Investment Management. We should simply buy when a monthly price is above its 10-month moving average and sell when it is below it, and move into cash instead.
I've already shown that this rule works for the All-Share index. My table shows how you would have done if you had applied it to two FTSE sectors - mining and IT - since December 1990.
How rules work
|10-month average||Buy and hold||10-month average||Buy and hold|
|Worst 12 months||-34.6||-62.6||-49.5||-84.0|
|Based on returns excluding dividends since December 1990|
For IT, the results are spectacular. Annual returns (excluding dividends) have been more than twice as high using the 10-month average rule as you'd have had under a simple buy-and-hold strategy. And volatility has been lower too. This is mainly because the rule got us out of the sector in November 2000 and, except for a couple of months, kept us out until the spring of 2003. We would therefore have avoided most of an 88 per cent loss. Granted, the rule missed the first leg of the 2003-04 bounce, but it got us into the sector for a 55 per cent gain between 2003 and 2007.
The rule has also worked for miners, giving us higher average returns, lower volatility and less bad worst-case losses. The rule would have got us out of the sector in July 2008 thus saving us from a 47 per cent loss, and would have sold last September, avoiding a 40 per cent loss.
However, although the ten month average rule for mining has beaten buy-and-hold, the difference between the two strategies isn't as spectacular as it is for IT.
To see why, think of the pattern of returns that would cause the rule to succeed or fail. If markets are streaky, so that gains lead to gains and losses to losses, the rule works well. If, however, markets are dippy - with prices rising gently, then dipping and then recovering - the rule fails. It gets us into shares near the top of the market and out of them just before recoveries. It so happens that the IT sector has been more streaky and less dippy on average than miners since 1990, and so the 10-month average rule has worked better for IT.
However, that phrase 'on average' is doing some work. The 10-month average rule underperformed buy-and-hold until around 2005. All of its outperformance has come in the last 10 years. This is because the sector used to be dippy but has since become streaky.
This might be no accident. In a world of secular stagnation and low interest rates, bubbles and trend-following become more likely. As Harvard University's Larry Summers has said: "It is only rational to recognise that low interest rates raise asset values and drive investors to take greater risks, making bubbles more likely." If this is the case, then investors should use the 10-month average rule, or something like it, to discipline themselves to get into and out of specific sectors.
For now, though, the rule is sending a very clear message. The mining sector is 24 per cent below its 10-month average which suggests that it is dangerous to buy. Granted, following the rule means we would miss out on the early stages of any recovery - which, who knows, might have already begun - but it protects us from downside risk. And in momentum-driven markets, this is useful.