PsyFi Search

Loading...

Tuesday, 9 June 2009

Momentum Trading Madness

Jungle Survival Strategies

When we used to skulk about in the rainforest as undernourished and over-predated apes, evolution figured out that a good strategy for survival was to assume that any trend was likely to continue, once it had established momentum. Of course this occasionally went wrong when the volcano god got an attack of indigestion but generally tomorrow was likely to be pretty much the same as today, the antelopes were always at the waterhole at the same time and wandering into the sabre-toothed tiger’s lair was never a good idea.

Unfortunately to make money safely on the stockmarket you’ve got to assume exactly the opposite of momentum: that what worked yesterday won’t work tomorrow and that the tiger's never where you think it is. Which is not the way our ape-evolved minds prefer to work.

Random Walking Stocks

Share prices follow a random walk, lurching about with all the finesse and awareness of a drunk on a boat in a storm. Many investors, though, operate assuming that shares have a memory and that this memory translates into momentum – that a share went up (or down) yesterday means it’s more likely to go up (or down) today. Quite clearly shares can’t both follow a random walk and genuinely demonstrate momentum. Or can they?

The random walk theory simply says that share prices meander about with no discernable short-term direction. It doesn’t say anything about the overall direction of share prices which, on average and over time, trend upwards with economic growth. Imagine our drunk determinedly lurching his way along the heaving deck in the direction of the Captain’s Bar. The direction of any single step will be completely random yet overall he’ll still be heading towards that final one for the road.

If this is correct then momentum trading shouldn’t work. Sure, on occasions the drunk will take several steps in the same direction, but eventually he’ll lurch off randomly and break the sequence. Overlaid on top of this, however, we have the standard mean regression effect acting on share prices.

Long term momentum

Now a moment’s thought shows us that mean regression implies momentum over time. If a share is under or overvalued and is to return to fair value then it must eventually move towards the mean – which implies directionality.

A study by ABN Amro has shown exactly this. In backtesting, investing each year in the top 20% of performers from the previous year produced quite stunning outperformance over the bottom 20%. In the UK, between 1957 and 2007, this would have produced a near 11%+ additional return. Applying the same approach to the same market going back to 1900 showed that £1 invested in the top 20% would have yielded a mightily impressive £4.5 million. Investing in the bottom 20% each year would have produced a rather less remarkable £111. It’s a finding which appears to apply to other markets, as well.

There are important caveats to this – rebalancing costs would be a significant drag on performance – perhaps up to 50% - and the strategy can break down for multiple years, with significant volatility. It’s not an approach for the faint hearted, but it's still an interesting finding.

Short term momentum

Unfortunately most people can’t follow strategies like this over long periods of time. Almost invariably people jump ship when it gets difficult, preferring short-term strategies based on momentum – buying stocks that have had good news or have a popular story to tell. So, most momentum trading techniques are short-term, looking for illusory trends in our drunkard’s walk. Just as we can find “trends” in the outcome of a coin tossing competition so we can in short-term share price movements.

In fact if there are any easily exploitable trends in short term momentum then they’ll rapidly get arbitraged away, such is the interest in such methods. So much investing is short-term – especially on the professional side – that it’s practically impossible for any sustained method of achieving short-term outperformance to survive. Long-term it’s different – the rewards for professional investors are so skewed in favour of the short-term that they virtually ignore anything longer than a few months. Of course our simian forebears were equally focused on the short term but they were rather more worried about the next meal rather than the next bonus check.

Mutual Fund Momentum

Mutual fund groups are great exponents of momentum – just as soon as any major trend is established in the market they’ll start to launch funds to exploit it, just in time to get going and suck in investor’s money before the trend fails. In my time I’ve seen this from Eastern European funds based on the collapse of communism to dotcom technical funds through to commercial property investments. Capital consuming, fee earning, momentum following bandwagons, the lot of them.

Although the tendency to jump on the nearest trend is all too evident in mutual fund managers' behaviour the major problem for investors is their inability to sustain a constant strategy. As the Dalbar studies have shown most investors fail to achieve even the underperformance offered by mutual funds. In greed they invest in hot funds at the end of bull runs and sell in panic at the tail end of bear crashes. Overall they managed a pathetic 3.7% a year between 1985 and 2004 while the market achieved nearly 12%. Even the mutual funds themselves made about 10%.

Ape-Think Means Following Momentum

Given that these deleterious strategies are so well known and the mean regression effect is so well understood why do momentum trading approaches persist in the markets? Well, the answer is almost certainly nothing to do with stocks or markets and everything to do with the way we’ve been evolved to think.

Mean regression is the staple effect behind most long-term investment trends. It most definitely is not the overriding effect in the short-term where it can go badly awry. However, investors who take a long term view and invest in stuff that’s undervalued, often for good reason, can pretty well guarantee an above average return - eventually. Unfortunately when you’re a poorly equipped, unarmed ape trying to scrape a living on the savannah you don’t really have the time to worry about the long-term – tomorrow is as good as it gets. So assuming that any given trend is likely to continue is a darn good quick processing option.

The simian mid-brain where most of our automatic thinking occurs – the easy short-cuts that get us through most days – works on momentum. It doesn’t do mean regression: you don’t assume the antelope will go somewhere else at watering hole time. So we need to seriously think, using human higher cognitive abilities, about these issues if we’re to make any sense of them. It’s genuinely hard work.

Think Like A Human, Not an Ape

So it’s much easier to not bother and assume that the trend-following automated processing that governs most of our lives is telling us something useful about stockmarket investing. Easier to find a tip, read someone else’s analysis or look for some trending mechanism like momentum to follow. Whether investors are looking for short-term momentum or simply chasing the latest exciting market fad being promoted by the mutual funds the underlying psychological triggers are exactly the same.

Occasionally, though, similar processes yield interesting results. The ABN Amro study quoted above, claiming to demonstrate that momentum trading works, does no such thing. It simply shows that mean regression effects, once established, tend to perpetuate themselves. and, no doubt, they often end up with hoards of short-term money flooding into them, just as the trend’s about to bend downwards. That’s the trouble with thinking like an ape in a mean regressing world.


Related Posts: Technical Analysis, Killed By Popularity, You Can't Trust The Experts With Your Investments

1 comment:

  1. Here are the results of test I did to see if trading momentum produced a statistical edge on the GBP/USD http://www.myforexdot.org.uk/momentum-trading.html

    It did, but this type of system seems to only work on a few markets such as the British Pound vs US Dollar. Other tests I've done say it's a disaster on other markets, most notably stocks and shares.

    ReplyDelete