Apparently it doesnt stand to empirical tests. But I'm open to contrary evidence
http://www.newyorkfed.org/research/staff_reports/sr42.pdf
---
Moderator's Notes
I was very curious how the article author determined that "head and shoulders trading is unprofitable," so I looked at his entry and exit strategy.
(1) The entry is a close below the neckline on the daily chart.
(2) The exit rules (for the Cliff's Notes version just read the bold):
So, in all the author's supposedly thorough research on how to trade a head and shoulders, he apparently missed Bulkowski's "Encyclopedia of Chart Patterns" which actually explains how to trade this pattern.
The entry and exit strategies used in this study are rather inept. They are missing all of the fundamental pieces required for a profitable trading methodology. Exit rules need to be designed specifically for the problem at hand, with the belief that history in some way will repeat itself. Quantitatively, we study the successful patterns and also the losers, designing entries and exits to take advantage of past behaviors. While many poo-poo this approach with the epithet "curve fitting," it actually does produce profits if done correctly, and the market behavior after each entry taken is characterized in order to determine if history, indeed, is in some way repeating itself.
So specifically for head and shoulders, to trade it profitably one must abide by these rules:
(1) Do NOT wait for a confirmed breakout. The proper entry is as price is NEARING breakout. Yes, it's an art. This is one thing the human mind does far better than computers: pattern recognition.
The author of this paper enters on a daily close below the neckline! This is quite late and it also convolutes the stop (see below).
(2) Stick to the profit target unless you have a quantitatively justifiable reason to get out (such as a time stop based on how long past winners have taken).
The author of the paper exits at the first trough unless he's less than 25% to the profit target; he exits at the second trough regardless of the profit target. These rules are arbitrary feel-good rules and are not based on past performance, as they should be.
(3) Use a proper stop. In this case, just above the neckline or the lower of the two troughs, whichever is higher.
The author uses a 1% stop according to his entry price. So sorry, but the market does not care where your entry is. The stop needs to be placed according to where H&S patterns fail, not how much money you might lose. THEN you adjust your size in order to risk 1% of your account, for example.
So, in summary, yet another paper using a lousy trading algorithm and thinking they've shown an aspect of TA doesn't work. Spare me.
PetaDollar
http://www.newyorkfed.org/research/staff_reports/sr42.pdf
---
Moderator's Notes
I was very curious how the article author determined that "head and shoulders trading is unprofitable," so I looked at his entry and exit strategy.
(1) The entry is a close below the neckline on the daily chart.
(2) The exit rules (for the Cliff's Notes version just read the bold):
So, in all the author's supposedly thorough research on how to trade a head and shoulders, he apparently missed Bulkowski's "Encyclopedia of Chart Patterns" which actually explains how to trade this pattern.
The entry and exit strategies used in this study are rather inept. They are missing all of the fundamental pieces required for a profitable trading methodology. Exit rules need to be designed specifically for the problem at hand, with the belief that history in some way will repeat itself. Quantitatively, we study the successful patterns and also the losers, designing entries and exits to take advantage of past behaviors. While many poo-poo this approach with the epithet "curve fitting," it actually does produce profits if done correctly, and the market behavior after each entry taken is characterized in order to determine if history, indeed, is in some way repeating itself.
So specifically for head and shoulders, to trade it profitably one must abide by these rules:
(1) Do NOT wait for a confirmed breakout. The proper entry is as price is NEARING breakout. Yes, it's an art. This is one thing the human mind does far better than computers: pattern recognition.
The author of this paper enters on a daily close below the neckline! This is quite late and it also convolutes the stop (see below).
(2) Stick to the profit target unless you have a quantitatively justifiable reason to get out (such as a time stop based on how long past winners have taken).
The author of the paper exits at the first trough unless he's less than 25% to the profit target; he exits at the second trough regardless of the profit target. These rules are arbitrary feel-good rules and are not based on past performance, as they should be.
(3) Use a proper stop. In this case, just above the neckline or the lower of the two troughs, whichever is higher.
The author uses a 1% stop according to his entry price. So sorry, but the market does not care where your entry is. The stop needs to be placed according to where H&S patterns fail, not how much money you might lose. THEN you adjust your size in order to risk 1% of your account, for example.
So, in summary, yet another paper using a lousy trading algorithm and thinking they've shown an aspect of TA doesn't work. Spare me.
PetaDollar
