Do candlestick patterns actually work?

[RowanC]

There are many trading concepts that I’m somewhat skeptical about. The merits of candlestick patterns certainly fall into that category.

Being a former intraday trader, these patterns seemed to be great at missing the entire move as in reality all the buying and selling that goes on in short term moves can’t be seen on a chart. So candlesticks always appear a little late to the party in my opinion.

However there are some patterns that as a trader appear to have merit. The hammer for instance is one such pattern that appears to look like it should conceivably identify a turning point of sorts. So I thought I would test that particular pattern out on a large universe of stocks on a daily timeframe.



I constructed the following criteria:

• ASX200 plus historical constituents back to 1992
• Stock is above its 200-day SMA
• Buy on the next open after we get the ‘hammer’ chart pattern
• Sell on the next open if the closing price is greater than the 5 period SMA
• Maximum of 10 positions at 10% each
• 0.08% commissions

So how do the results look? Actually pretty good.



It’s fair to say that I was stunned that not only did this strategy turn a profit, but the equity curve looked reasonable.

However when we dive into the metrics there does seem to be a few issues that we have to address.

CAGR 13.33%

MaxDD -44.12%

Sharpe Ratio 0.44

Winners 53.05%

Just looking at these numbers we can clearly see that this isn’t an amazing strategy. That said this is not optimized on any level and is purely for the purposes of seeing if there is any merit to candlestick patterns.

For a point of comparison the XAOA has returned 8.9% pa since 1990 so in terms of an absolute return then our simple hammer trading strategy isn’t looking so bad.

If there are any particular candlestick patterns anyone wants tested, please leave a comment below.

rowancrosby.com

 

[peter2]

Nice bit of work. I'd be interested to see the results of a weekly bar.

I use the weekly charts more than the daily charts and I look for three candle patterns. I consider the bars as price bars rather than candlestick formations with pretty names.

I know you'll ask about the three.
1. Outside reversal bar (key reversal)
2. Pin bar like what you tested earlier.
3. High close doji. Usually a two bar pattern (can be more). Comprises a doji followed by a bullish bar that closes above the high of the doji. The Morning Star pattern is one example of this concept.

Price bars by themselves don't test well. However at certain chart locations they shine. eg A hammer at support, previous swing high or after an abc correction.

Do they work? No, but they are objective setups that can be used as part of a profitable strategy.

 

[peter2]

Forgot to ask if you tested the hammer as a five bar pattern, three bar pattern or just the one bar pattern. The one bar "hammer" pattern is also a one bar "hanging man" pattern. Location and prior bars mean a lot more than the one bar.

 

[RowanC]

Forgot to ask if you tested the hammer as a five bar pattern, three bar pattern or just the one bar pattern. The one bar "hammer" pattern is also a one bar "hanging man" pattern. Location and prior bars mean a lot more than the one bar.

Hey Peter

Thanks for the info. I'm by no means an expert on the patterns so it's great to get some clarification.

In terms of the pattern that I used, the requirement was that the previous bar (prior to what I called the hammer) was higher. A higher high and higher low. So effectively a downtrend - albeit a short one.

It's interesting that you mentioned to use them in conjunction with support, as from a purely anecdotal perspective that appears to be effective so I'll do a test of that in the future along with some of the other patterns you mentioned.

I'll run this test again, on weekly data for you also and put the results in the comments.[/QUOTE]

 

[RowanC]

Nice bit of work. I'd be interested to see the results of a weekly bar.

I use the weekly charts more than the daily charts and I look for three candle patterns. I consider the bars as price bars rather than candlestick formations with pretty names.

I know you'll ask about the three.
1. Outside reversal bar (key reversal)
2. Pin bar like what you tested earlier.
3. High close doji. Usually a two bar pattern (can be more). Comprises a doji followed by a bullish bar that closes above the high of the doji. The Morning Star pattern is one example of this concept.

Price bars by themselves don't test well. However at certain chart locations they shine. eg A hammer at support, previous swing high or after an abc correction.

Do they work? No, but they are objective setups that can be used as part of a profitable strategy.

Hi Peter

I reran the analysis this time on a weekly time frame using my same conditions, however I adjusted the requirement of the Close being above the 200-day SMA, to it being greater than the 40-period SMA.

On the ASX200 plus historical constituents the results came in around breakeven in terms of the CAGR. However running it again on the All Ordinaries plus historical constituents to get more data points, showed a 7.26% CAGR which was interesting.

So perhaps there is more value coming from smaller stocks, which can often be the case.

Again as discussed this is by no means a robust test, but it does show that there is potentially a bit of value there over a long period of time.

Cheers
Rowan.

 

[peter2]

Your back-test using the daily pattern was much more robust than I'd realised, so I wasn't surprised that the weekly didn't show a significant improvement. Thanks.

 

[tech/a]

You have chosen Candlesticks as your topic
I would add VSA point and figure all forms of charting from kagi to bar charts infact technical analysis it self.

I see your questioning in similar posts by many others.
Isolating everything from single bar set ups to multiple bar setups like
Rectangles and triangles.

After over 20 yrs of workin in T/A and trading using it
I think you and the many are looking at analysis in a very narrow way.

I came to an epiphany many years ago when I realized all analysis

Indicates and anticipates possible and probable moves in any chosen timeframe.

You and others are looking for a definitive answer in singular patterns

You and others are very disappointed at what you see
I was at onetime with you!!

But then I realized that CONTEXT was the missing link.
All patterns singular or in combination need to be read in context of any chart and timeframe.

Once you can group the technical indicators and read them in context to any chart you'll find that they will be very accurate in determining the 3 things we really need to know from technical analysis

Is the chart flat?
Is the chart changing course
Is the chart stopping it's current direction

Once mastered put these inputs into you methodology and then ask your question----the answer I expect---- after the journey will be clear.

 

[RowanC]

You have chosen Candlesticks as your topic
I would add VSA point and figure all forms of charting from kagi to bar charts infact technical analysis it self.

I see your questioning in similar posts by many others.
Isolating everything from single bar set ups to multiple bar setups like
Rectangles and triangles.

After over 20 yrs of workin in T/A and trading using it
I think you and the many are looking at analysis in a very narrow way.

I came to an epiphany many years ago when I realized all analysis

Indicates and anticipates possible and probable moves in any chosen timeframe.

You and others are looking for a definitive answer in singular patterns

You and others are very disappointed at what you see
I was at onetime with you!!

But then I realized that CONTEXT was the missing link.
All patterns singular or in combination need to be read in context of any chart and timeframe.

Once you can group the technical indicators and read them in context to any chart you'll find that they will be very accurate in determining the 3 things we really need to know from technical analysis

Is the chart flat?
Is the chart changing course
Is the chart stopping it's current direction

Once mastered put these inputs into you methodology and then ask your question----the answer I expect---- after the journey will be clear.

Thanks for your thoughts tech/a.

While this test was certainly very simplistic, I'll be continuing to build on this in further posts.

Cheers