Many of you are worried about the high valuations of some US tech stocks such as Tesla. There’s not much point debating whether they really are overvalued, because even overvalued shares can continue rising just as undervalued ones can continue falling. Instead, we should ask how best to manage the risks here.
And they are big risks, because when overvalued shares fall they can fall a lot. When the tech bubble of the late 1990s deflated the Nasdaq composite index lost 75 per cent and did not return to its February 2000 peak until late 2014. The UK’s experience was even worse: the FTSE IT sector lost 95 per cent when the bubble burst and even today is only half its 2000 peak.
Which poses the question: can we protect ourselves from such catastrophic losses without missing out too much on the upside?
In theory, yes. Mebane Faber at Cambria Investment Management has proposed a simple rule – that we should be in shares when prices are above their 10-month (or 200-day) moving average and get out when they fall below it. Such a rule allows us to ride bubbles as they inflate while jumping off before they fully deflate – albeit not at the top of the market.
He has shown that such a rule would have worked well for investors in the S&P 500. It has also worked well for emerging markets and for bubble-prone UK sectors such as mining and IT. But does it work for US tech stocks?
To answer this, I applied the rule to the Nasdaq composite index since 1999, being in the market if the index was above its 10-month average at the end of the month and out of it if it was below. I assume that if we were out of the market we earned zero return – that is, today’s interest rate rather than past rates.
If we look at returns alone, the rule has worked only slightly. Since December 1999 it would have given us an annualised capital gain of just under 6.3 per cent, compared with 6.1 per cent on the buy-and-hold strategy.
Where the rule has done well, though, is in reducing risk. The standard deviation of returns on the 10-month rule has been just two-thirds that of the buy-and-hold strategy. Thanks to this, its Sharpe ratio has been more than 50 per cent better.
Better still, the rule would have protected us from really big losses. The worst annual loss with the rule was less than 40 per cent, compared with almost 60 per cent for the buy-and-hold strategy.
Such protection happens because the rule gets us out of a falling market. It got us out of the Nasdaq between October 2000 and December 2001 thereby avoiding a 40 per cent loss. And it got us out from February 2008 to May 2009, thus avoiding a 20 per cent loss.
The case for such a rule, therefore, seems clear.
Except that it might not be.
Imagine the market were to follow a saw-toothed pattern, with moderate falls alternating with rises. The 10-month rule would then be a disaster. It would have us in the market just before falls and get us out before recoveries.
In fact, you don’t have to imagine this. Had you followed the 10-month rule between 1985 and 1999 you would have significantly underperformed buy-and-hold.
So why the difference? It’s about market ecology – which species of investor inhabits the market environment.
If the market is dominated by value investors then the 10-month rule will fail. Falls in the market will attract buyers, so getting out before these buyers emerge is daft.
If, however, momentum investors dominate then rising prices lead to further rises and falls to further falls, in which case selling after a moderate fall protects us from worse losses.
Since 1999 we’ve seen two big momentum-driven falls – those of 2000-03 and 2008-09. In protecting us from a lot of these losses, the 10-month rule enhanced long-term returns.
Which isn’t to say that it works perfectly. It would have got us out of the Nasdaq at the end of March, thereby missing April’s 15 per cent jump – although it would have got us in for the subsequent rise in prices.
The question, then, is: what do you want a strategy to do? If you want to maximise potential upside, the 10-month rule is not for you. If, however, you want to reduce the chance of a horrible once-in-a-decade loss, then perhaps it is.
