Candlesticks not profitable Massey University Research

Do Candlesticks work or not?


  • Total voters
    135
................. It boils down to the ol' chestnut of Mechanical Trading V's Discretionary Trading and why - in my humble opinion - the latter will always prevail over the former in the long run................
Tim.


tim,

unlesss your "mechanical" system is built from your "discretionary" reasoning - if you see what i mean :rolleyes:

good trading

jon
 
The most bizarre aspect surely must be using a metric such as assessing the profitability after X bars.

It surely sounds like an arbitrary and bizarre choice. But what about actual candle stick traders? What do they use? Do they maybe use fitted exits? or trailing stops? or gut feelings?

The argument is as follows (in the study): candles sticks should be short term patterns. If that is true and they work they should be a good market timing method for a period of days in the order of the period in the pattern formation.

This method of exit (X number of days) appears good to establish whether the formations provide timing quality. Do not confuse that with a method to maximize profits or get a better equity performance. That was not the subject of the study it seems.

The issue is whether candle sticks are efficient market timing indicators. If, like someone else insisted before, you need to know the direction of the trend to use candle sticks then any profits obtained by using them may be pure chance.

Regardless the poor quality of the study I think they illustrated what traders know for a long time now, that is the fact that candle sticks is only a way to graph data, as someone else pointed out, I think it was Tim.

Alex
 
"Basing trading decisions solely on these techniques does not seem sensible but we cannot rule out the possibility that they compliment some other market timing techniques."

NO SH*T SHERLOCK!
 
It boils down to the ol' chestnut of Mechanical Trading V's Discretionary Trading and why - in my humble opinion - the latter will always prevail over the former in the long run.
Tim.

Correct, one explanation for that is very very simple: discretionary trading is what makes prices move at points where valuations are difficult such as market bottoms or tops.

Discretionary traders make profits by moving markets. A small percentage of mechanical traders make money by trading the noise discretionary traders make. Most of mechanical traders become the victims of discretionary traders.

Alex
 
:LOL:
The Massey University Finance Dept. carried out research by Ben R. Marshall, Martin R. Young and Lawrence C. Rose, Dept. of Finance established from their research that candlesticks do not work and are not profitable, this research was released 16 April 2007.

We investigate the profitability of the quantitative market timing technique of candlestick technical analysis in the U.S. equity market. Despite being used for centuries in Japan and now having a wide following amongst market practitioners globally, there is little research documenting its profitability or otherwise. We find that these strategies are not generally profitable when applied to large U.S. stocks. Basing trading decisions solely on these techniques does not seem sensible but we cannot rule out the possibility that they compliment some other market timing techniques.

Download research link http://ssrn.com/abstract=980583


:LOL: Thank God, or praise be for academia. Have they concluded its random then ?

Finance department , hmm wrong department to be studying "trading" techniques ?

Must try harder, or not... :)
 
The value in testing elements of trading in this:

Traders search for certainty and in that search they add elements that might give it to them. They try to use other markets. They look at internals. They look at volume. They see patterns in anything from candlesticks to tea leaves and in some cases the moon. But ... they are often not at all rigorous in checking if real value is added. Because of that traders seem almost superstitious at times.

The report suggests situations in which these traditional candlestick patterns are not adding value (most).

If you see methodological flaws in the report then it would be interesting to explain them and how they should be corrected.
 
The value in testing elements of trading in this:

Traders search for certainty and in that search they add elements that might give it to them.

Certainly this is true and the result is something equivalent to Multicollinearity in regression.

Essentially, most traders introduce more elements to reduce the number of losing trades of their system. This is easy to do but the various elements interact in peculiar way depending on factors that may not arise again in the future.

The right approach is to try to introduce elements that increase the number of winners. This is difficult to do.

Now, think about it: does the market repay you for doing easy or difficult things with your system?

Alex
 
tim,
unlesss your "mechanical" system is built from your "discretionary" reasoning - if you see what i mean :rolleyes:

I do indeed Jon, although I suspect you'll need to be a techie genius to write the program based around discretionary criteria!
--------------------------------------------------------------------------------------------------------

The alleged ineffectiveness of candlestick patterns is easy to demonstrate in ShareScope. Recently, they upgraded the software and announced with great glee a new facility enabling subscribers to flag up 12 or so of the most common candlestick patterns. Users can edit which candlestick formations they want to be alerted to so that, for example, the software only searches for bullish engulfing candles. In the chart attached, I've selected the 'show all' option, i.e. all bullish and all bearish patterns.

The stock (SDR) is clearly in a strong uptrend, having risen from around £9.00 to high of £14.00 in nine months. Given this, it is surprising then that there isn't one single bullish pattern alerting a potential long trade. So, anyone looking for prospective long trades would not have had this stock on their radar because no bullish signals were generated. However, any contrarians looking for shorting opportunities would have been alerted to SDR no less than 10 times since Oct 06. With the possible exception of the signal generated in late Feb, the other 9 signals - if acted upon - would probably have resulted in losing trades for the swing or position trader. Not good, not good at all.

There are two obvious conclusions to be drawn from the chart:-
1. that ShareScope's software isn't too hot and needs re-programming!Or . . .
2. that Candlestick patterns don't work.
A quick trawl through the FTSE 100 and NAZ 100 reveal a different type of pattern evident in both markets. When stocks are trending up, bearish patterns are generated and when they are trending down, bullish patterns are generated. So, just as the chart of SDR shows, using these patterns to enter trades is likely to disappoint. However, they may be of some use in exiting part or all of a position or at least in alerting a trader that possible change lies ahead, in the short term at least.

I don't use this facility within ShareScope because it has no value at all for me. It's an unnecessary and unwelcome filter that prevents me from thinking about the forces at play that cause price to move the way that it does. To this end, all candles have some meaning and value and need to be assessed in context. However, in isolation, individual candles are usually pretty meaningless and are more likely to confuse than they are to enlighten.
Tim.
 

Attachments

  • Candlesticks_1_10_06_07.jpg
    Candlesticks_1_10_06_07.jpg
    101 KB · Views: 308
Candlesticks depict traders actions and behaviour in a past trading session or sessions, they provide with a visual picture of trading action which has ocurred. How can we as traders use candles to predict prices in the future?
 
Massey University invitation to take part in Candlesticks debate

I started a thread about your research two days a ago and so far I have had around 550 viewings this forum thread is proving very popular among traders. I have also set up a poll about do candlesticks work or not and 60% say yes 40% say no.

Steve Nisons has sent his response to your research which is on the thread, Steve disagrees with your research. Will you please take part in this debate by posting your views on this thread we traders will very much appreciate your participation. This debate is of very importance around the globe judging by the response which are coming in from around the world.

The objective of this debate is not to prove who is wrong or right but to achieve greater understanding about candlesticks and how effectively we as traders can use candlesticks to make our trading decisions.

We look forward to your reponse on this forum which is highly active and particpation will give an extra characteristic and an informative dimension.

http://www.trade2win.com/boards/showthread.php?t=25220

I appreciate your quick response to my emails and value and respect your knowledge and the time and effort you have allocated to candlesticks research. Candlesticks have developed strong following in the Western nations and your knowlege and research has openned a new avenue in this puzzle.
 
I do indeed Jon, although I suspect you'll need to be a techie genius to write the program based around discretionary criteria!
--------------------------------------------------------------------------------------------------------

The alleged ineffectiveness of candlestick patterns is easy to demonstrate in ShareScope. Recently, they upgraded the software and announced with great glee a new facility enabling subscribers to flag up 12 or so of the most common candlestick patterns. Users can edit which candlestick formations they want to be alerted to so that, for example, the software only searches for bullish engulfing candles. In the chart attached, I've selected the 'show all' option, i.e. all bullish and all bearish patterns.

The stock (SDR) is clearly in a strong uptrend, having risen from around £9.00 to high of £14.00 in nine months. Given this, it is surprising then that there isn't one single bullish pattern alerting a potential long trade. So, anyone looking for prospective long trades would not have had this stock on their radar because no bullish signals were generated. However, any contrarians looking for shorting opportunities would have been alerted to SDR no less than 10 times since Oct 06. With the possible exception of the signal generated in late Feb, the other 9 signals - if acted upon - would probably have resulted in losing trades for the swing or position trader. Not good, not good at all.

There are two obvious conclusions to be drawn from the chart:-
1. that ShareScope's software isn't too hot and needs re-programming!Or . . .
2. that Candlestick patterns don't work.
A quick trawl through the FTSE 100 and NAZ 100 reveal a different type of pattern evident in both markets. When stocks are trending up, bearish patterns are generated and when they are trending down, bullish patterns are generated. So, just as the chart of SDR shows, using these patterns to enter trades is likely to disappoint. However, they may be of some use in exiting part or all of a position or at least in alerting a trader that possible change lies ahead, in the short term at least.

I don't use this facility within ShareScope because it has no value at all for me. It's an unnecessary and unwelcome filter that prevents me from thinking about the forces at play that cause price to move the way that it does. To this end, all candles have some meaning and value and need to be assessed in context. However, in isolation, individual candles are usually pretty meaningless and are more likely to confuse than they are to enlighten.
Tim.

Tim

au contraire, mon brave :)

Since most candlestick patterns are reversal patterns with relatively few continuation patterns it's no surprise to see them appearing in SDR. Also, in candlestick terms, "reversal" means a change to the existing trend which means going flat counts and not just full reversals. As I see it the candles worked quite well on this basis in the SDR chart you posted (not that engulfing patterns are among my favourites) - they may not, in most cases, have given reliable short entries, but they would have given you some good exits with re-entry lower down as you rode the trend.

good trading

jon
 
Tim

au contraire, mon brave :)

Since most candlestick patterns are reversal patterns with relatively few continuation patterns it's no surprise to see them appearing in SDR. Also, in candlestick terms, "reversal" means a change to the existing trend which means going flat counts and not just full reversals. As I see it the candles worked quite well on this basis in the SDR chart you posted (not that engulfing patterns are among my favourites) - they may not, in most cases, have given reliable short entries, but they would have given you some good exits with re-entry lower down as you rode the trend.

good trading

jon
A very good point barjon.

As barjon points out in candlestick terms a reversal pattern only means a change of the trend over the short term. A change is either a complete reversal or a going flat. Does anyone have research on whether this meaning holds any validity? For example, does a shooting star pattern truly indicate whether the trend flattens or changes any more accurately than mere chance in a statistcal sense?

Candlesticks are not buy and sell signals at all. They are graphical representations of what occurred during a trading session. Testing them as buy and sell signals is like testing to see if apples taste like oranges. You are testing for something that just isn't there.

My two cents anyway......

Cheers,
PKFFW
 
Candlesticks are not buy and sell signals at all. They are graphical representations of what occurred during a trading session.

Actually, nothing can be a trading signal. All indicators tell you what happened in the past. Nothing can tell you what will happen in the future with certainty.

So, saying that "Candlesticks are not buy and sell signals at all" maybe received as a red herring by some people. Better I suggest the following: Candlesticks, too, are not buy and sell signals at all .

Technical traders base decision on price and volume action and on anything relating to that, including candle sticks. The issue is whether candle sticks can provide high profitability entry/exit points. My past work based on traditional candle stick formations says a flat no. However there is a whole class of unknown candle stick formations with remarkable performance statistics.

Alex
 
Actually, as usual you will not like this...

But the research conclusion coming out of Massey University is correct.

Candles do not work.

Candles are inanimate objects created to represent price tracks within a given period. Of themselves they tell you very little in the way they are displayed.

What tells you a lot is how you view them and what you are able to percieve by having an understanding of what they represent and why.

I am not going to go too deeply into this.

.:LOL:
 
Last edited by a moderator:


Candles are inanimate objects created to represent price tracks within a given period. Of themselves they tell you very little in the way they are displayed.

What tells you a lot is how you view them and what you are able to percieve by having an understanding of what they represent and why.


Hi Socrates, glad you contributed to this. I have no problems with the research methodology, or indeed its findings. They formed a hypothesis, and tested it. Obviously they could have saved themselves a lot of time trouble and expense, as the outcome was obvious !

However, If I remember correctly, the study didnt include the use of volume, Id be genuinely interested to hear your thoughts on that, for example do you feel that certain patterns may have proved to be statistically significant if perhaps volume had been included in the analysis ?

If I where a vendor, peddling a product that was candlestick based, I think Id try to defend my product, by pulling the "lack of volume" argument, and Im genuinely surprised that to date no one has. I'd be interested to hear your take on this.

regards
zu
 
Actually, nothing can be a trading signal. All indicators tell you what happened in the past. Nothing can tell you what will happen in the future with certainty.

So, saying that "Candlesticks are not buy and sell signals at all" maybe received as a red herring by some people. Better I suggest the following: Candlesticks, too, are not buy and sell signals at all .

Technical traders base decision on price and volume action and on anything relating to that, including candle sticks. The issue is whether candle sticks can provide high profitability entry/exit points. My past work based on traditional candle stick formations says a flat no. However there is a whole class of unknown candle stick formations with remarkable performance statistics.

Alex
Alex,

You make my point exactly. All successful traders I know and have heard of make buy and sell signals based on a confluence of events. Taking any one aspect or event singularly and attempting to test whether that makes a valid entry point is nonsensical.

In candelestick theory, it is suggested the patterns indicate a certain propensity for something to happen. For example; a reversal pattern, it is suggested, indicates it is likely for the current trend to either flatten out or reverse altogether. That is all it indicates.

Some traders may base their buy and sell signals on that information coupled with other information occurring at the same time. To take the pattern alone and test whether you would make money buying or selling based solely on that pattern is in effect testing for something that the candlestick theorists make no claim to. It would be more valid to test whether what they claim has any validity. In this case, whether or not the reversal pattern in question indicates a flattening or reversing of the current trend any more so than mere chance would. I would be interested in any such research and would be happy to be shown the truth either way.

You mention other candlestick formations with remarkable performance statistics. I would be interested to hear more on that subject too if you care to share.

Cheers,
PKFFW
 
Candles are inanimate objects created to represent price tracks within a given period. Of themselves they tell you very little in the way they are displayed.

What tells you a lot is how you view them and what you are able to percieve by having an understanding of what they represent and why.
Valid point, one I believe Spitlink(apologies if it was someone else) made a couple of pages ago.

- deleted

I edited out the content quoted, but I think PKFFW's comments on it below are very valid so I am leaving them for the moment.

jon
This is a prime example of why you draw so much criticism on these boards Socrates.

Here is a thread that has run 4 pages without an insult. You come on here and rather than simply stating your opionion and then leaving having said your piece or possibly entering into a discourse, you feel the need to post a thinly veiled insult. Why is that?

I have said on many occassions that you do have interesting and thought provoking things to say. Why do you find it so impossible to simply state them and leave it at that? And please do not give any long winded speeches about how everyone attacks you. In this thread no one has mentioned you, no one has disagreed with you, no one has insulted you. Your very first post has been to make a valid comment and then back it up with an insult.

If you get attacked in this thread you have absolutely no one to blame but yourself.

Cheers,
PKFFW
 
Did they include research on the heikin-ashi candle sticks?

Those highlight (& hold) trends, and would be interesting if they compared analysis between conventional candle sticks with the heikin-ashi candles.
 
Alex,

You make my point exactly. All successful traders I know and have heard of make buy and sell signals based on a confluence of events. Taking any one aspect or event singularly and attempting to test whether that makes a valid entry point is nonsensical.

Exactly. This is what succesful traders do as you say. You need a combination of techniques each adding to total value. The question is whether traditional candle sticks can provide such value. The answer is no, unfortunately, and one reason for that they have been extensively advertized and used.

This is the key point: if a method is widely used, the law of large numbers comes into play and the performance revertes to zero mean. People resist to accept this. But it is a fact of statistics. You need to use methods with an edge, methods not widely used.

You mention other candlestick formations with remarkable performance statistics. I would be interested to hear more on that subject too if you care to share.
PKFFW

A few years back and after attending a seminar on candle sticks, I realized that those formations are combinations of the open, high, low and close of bars plus, in certain cases, volume. I then came across APS Automatic Pattern Search, a program that finds price patterns based on your own criteria. Now, there is no essential difference between price patterns and candle sticks as far as formations go. The important difference is that there are thousands, literally, of price pattern formations that can be thought as candle stick formations but they are not known, usede or advertised.

Check out this link for an article on price patterns:

http://www.tradingpatterns.com/About_Us/articles/article2/article2.html

I use price patterns as part of a trading method that involves several other indicators as well as discretionary filtering of signals.

Alex
 
Top