Patrick Burns TA Test

[Trader333]

The most interesting part of the article, (for me), is the part which effectively says that chart analysis is no more effective than random decision making, although it was for EOD and not intra-day trading:

It may be, but but there are substantial doubts about whether chartists can really offer useful guides to future market movements. In a recent test in the Financial Times, Patrick Burns, a statistician, challenged chartists to examine 100 price series, each with 500 daily closing prices. Four possible extensions of the series were offered. One was the actual result and the other three were random. Burns asked analysts to pinpoint the true series.

What was the result?

Pretty poor. There were four possible answers for each series, so a purely random success rate would have been 25 per cent. The chartists who replied to the FT's challenge scored a hit rate of 26.3 per cent and no analyst did particularly well.

 

[qaza]

The most interesting part of the article, (for me), is the part which effectively says that chart analysis is no more effective than random decision making, although it was for EOD and not intra-day trading:

Anybody know where to get the details/article on the test ?

 

[options]

It would be a pretty good exercise if we could have that test on here. As long as the volume came with the charts.

 

[qaza]

options,

the results would be the same

 

[Mayfly]

A good point, Options - unfortunately it didn't come with volume and the series consisted of the close only.

Here's what I posted about the test on another site:

"Here's something that dropped through my "letter" box this morning and that I thought may be of interest to some people here.

http://www.burns-stat.com/

Mr Burns and his team are agnostic about Technical Analysis and they're sceptical about chartists abilities to distinguish between charts using real data and randomly generated data.

What they've done is to take the charts of 100 instruments using actual or real data and changed the scale and the time axes to conceal what the instruments are. They've then added four outcomes or "forecasts" to the data for each instrument for the next 50 days. One of these "forecasts" is based on actual or real data, the others are based on computer generated random data. The challenge is to identify the charts that are based on the actual or real data from the charts that are based the randomly generated data.

That's the gist of it. Sounds easy peasey to me......." :cheesy:

To cut a long story short: I declined the "challenge"

HTH

Cheers

Mayfly

 

[Sharky]

The article published in the FT about the tests is here:
http://search.ft.com/search/article...ore&vsc_publicationGroups=TOPWFT&searchCat=-1

And Patrick Burns website who conducted the test is here:
http://www.burns-stat.com/

And the test itself is here:
http://www.burns-stat.com/pages/techanchal.html

 

[Sharky]

Well done Mayfly, you got there first!

 

[TheBramble]

The most interesting part of the article, (for me), is the part which effectively says that chart analysis is no more effective than random decision making

Welles Wilder (of RSI fame) developed an interesting model a few years ago based on this approach.

He showed that providing you had correct Money and Risk Management in place, it is possible to randomly select stocks and make money.

No charts, no fundamentals.

It's not an approach I'm going to adopt, but based on some past 'methods' and 'systems' I've used, it's possibly got some merit.

 

[oatman]

Burns' test was useless as it was a "what happened next" from a neutral position rather than give you the opportunity to say where you would enter or exit a position.

 

[chump]

Reading Burns I have only one comment to make. Mr D are your skills available for hire ? 89% hit rate !!!!!


Cheers

 

[Bluebelle]

Oatman

That's absolutely right. I had a look at it in good faith, but it seemed to me that it was set up in a vacuum to give this guy's theory the best chance of working so I didn't touch it. In real life, no-one in their right mind would trade only on the basis of that sort of data and to draw the conclusions he did on the basis of that sort of spurious 'research' was laughable.

 

[Sharky]

Thread split from Darren Winters thread and moved to Technical Analysis forum.