SwingTrade Member Forum

[Rawnoc]

Chart hits, or misses?
By Philip Coggan
Published: September 5 2003 11:56 / Last Updated: September 5 2003 11:56


A challenge has been issued. Technical analysts, those who believe that future price movements can be divined from the patterns in past data, are being asked to prove their skills.


Patrick Burns of Burns Statistics has set a test on his website (www.burns-stat.com/test/pages/techanchal.html) for the chart gurus. He says that "technical analysts who use it to manage other people's money or to tell people how to manage their own money should feel honour-bound to participate publicly." The test may clarify the empirical basis for a technique that has many adherents and, rather like a religion, many different sects. Some chartists believe in short-term patterns, some in moving averages, some in longer term waves that can affect the markets over decades or, indeed, centuries.

They all believe that, by studying these patterns, investors can find clues that will enable their portfolios to outperform the market. However, most academics are dubious about their claims and fear that the many signals generated by technical models may simply saddle investors with higher dealing costs, thereby reducing their returns.

Proponents of technical analysis can point to two encouraging trends. The first is that believers in the "efficient market hypothesis" have been on the retreat. This school believed that prices were set with regard to all publicly available information, including past price patterns, so that the study of such patterns could not be profitable.

However, events such as the dotcom bubble have cast doubt on the efficient market theory. The intellectual trend has been running in favour of behavioural finance, which postulates that investors have psychological biases that cause them to behave irrationally; such biases can be exploited by others for profit.

The second trend is the growth of the momentum school of investing. Such investors tend to latch on to a change in market direction and by their actions, help to perpetuate that trend. Since much technical analysis is devoted to discovering trends, it is conceivable that this could offer exciting opportunities for chart followers.

One problem is that the evidence in favour of technical analysis can be highly anecdotal. After writing on the subject earlier in the year, I received many e-mails from people who either claimed, via their own trading, to make money from technical analysis or who knew others who did so.

Alas, one cannot really prove the validity of an approach using anecdotes. Those who had lost money through technical analysis were unlikely to respond.

There are two potential methods for examining the success rate of technical analysis. One is to analyse the records of all traders who have used it. Getting such information is quite difficult. The second is to test technical analysis "rules" or well publicised chart patterns, such as the double top, to see if they have predictive power.

There have been some studies done on this issue. Carol Osler of the US Federal Reserve conducted a study of head and shoulders patterns in the equity market over 31 years and found that, on balance, trading on the basis of such patterns was unprofitable.

However, the same Ms Osler found evidence that, in the foreign exchange markets, those who used support and resistance levels were quite successful in predicting intra-day trend interruptions. (It is worth noting, however, that it would be difficult for the retail investor to make money out of intra-day currency trading.)

So Burns has come up with his own method for analysing the claims of technical analysis. On his website, he has posted 100 series of prices. Each series comprises 500 daily closing prices of developed market equities. After each series, the contestant will be asked to choose between four potential continuations of the series, lasting 50 days. One of those options will represent the actual price movements of the shares; the other three will be randomly generated. The aim is to pick the real data.

The data have been disguised by starting all the series with a price of 10 and the start dates were chosen at random. There was no selection involved in the series continuations, nor were there any restrictions on the random series.

Burns says that contestants will not be expected to give answers for all 100 series but the more answers they give, the more likely it will be that they are demonstrating skill. Those who participate publicly will be required to submit at least 10 answers.

Burns adds that "a positive result for this test is only circumstantial evidence - not proof - that a technique can make money. However, being able to pick out the true prices seems like strong evidence for the usefulness of the technique."

It will be interesting to see if many technical analysts take up the Burns test. Many will probably dismiss it as being too artificial to catch the subtleties of their skills. Responses to my earlier article on the subject tended to suggest that the technique could not be judged by the success of defined rules or patterns. Instead the good chartist had to be able to interpret those patterns.

But Burns says that "we do not deny that some people probably have good intuition about the market. (But) a person's advice should not be dressed up with technical analysis if all of its value derives from intuition and not the technical analysis[/U]."

FT Money will update readers on how many technical analysts choose to take the Burns test and how many of them are successful. To misquote English football hooligans, go on, have a go if you think you're smart enough.