Predictive Power of Bearish Chart Patterns
This is a little, pseudo-study I did on SPY (SPX) analysing bearish chart patterns and whether or not there was a statistically significant relationship between bearish chart patterns and sell-offs and if so, to what degree?
Most of the literature says that chart patterns and technical analysis are not useful and that chart patterns are the circumstances of randomness. Most research argues that you cannot dictate what a stock will do based on the candlestick charts and patterns, as these only offer retro-spective perspectives in the stock and are not able to be used prospectively.
I, as someone with formal education in statistics and also as a full time trader, am torn between these analyses. On one hand, I can see where the researchers are coming from, on the other hand, I know it can’t be 100% true because many people are successful trading these strategies. While my personal strategy relies primarily on math, I would be lying if I said that I did not let chart patterns influence my decisions to some extent.
Also, the majority of people conducting these studies are not themselves traders. I, as someone who published and familiar with conducting research studies, figured meh. Why not get it straight from the horses mouth and do this myself. So alas, here we go!
I will present this as academically and rigorously as possible!
Procedure:
I looked at ONLY bearish chart patterns, particularly the patterns: head and shoulders, double top and triple tops. This was conducted ONLY using the 1 hour time frame, as the 5 minute time-frame would have been too labour intensive.
Additionally, I had included bear flags in the bearish chart patterns, however bear flags usually accompanied head and shoulders and double tops, which were more visible first and thus, I ended up not having any cases of bear flags to include.
The time frame was 1 hour and I included just over 2 years of data, from 2020 through till Friday.
The data was collected using TradingView’s platform and analyzed using SPSS v25.
Definition of Bearish Chart Pattern
I was somewhat lax with my interpretation of double tops and head and shoulders; however, the major inclusion criteria is that they had to be clearly interpretable as the bearish chart pattern. If I had to look and analyze it in-depth with lines, etc., then it was not included. It did not need to be perfect, it could be lopsided, it just needed to be easily recognizable as the pattern. However, Heikin Ashi candles were used to easily identify the patterns and verified via line charts.
Results
Between January 2020 and April 2022, on the 1 hour time frame, there were a total of 24 clearly observable bearish chart patterns. The breakdown is as follows:
Double Top: 12 incidents (50%)
Head and Shoulders: 9 incidents (37.5%)
Triple Top: 3 incidents (12.5%).
The success rate of all patterns was 83.3%.
The most successful chart pattern was a triple top. Of the three documented incidents, the success rate was 100%. The next was head and shoulders with a success rate of 89% (8 passes and 1 fail). The least successful pattern was the double top with a success rate of 75% (9 successes and 3 fails).
The results, and this is the KEY point, WERE in fact statistically significant via analysis through T-Test with a P value of 0.001!!!!! This means that these results cannot be attributable to chance or randomness and in fact, have statistically significant meaning.
Out of all the patterns, the average sell off that accompanied it was a drop of 10 points (standard deviation 11.27). Of the instances that failed, the average rise in price was 9.72 points (standard deviation of 4.66).
Below is a list of the average point sell off by chart pattern when successful:
Head and Shoulders: -12.33 (SD = 9.58)
Double Top: -8.59 (SD = 14.03)
Triple Top: -9.06 (SD = 5.56)
Conclusion / Limitations:
Please note, that this is far from what I would deem scholarly research. The problem with this type of research is trading is highly subjective and in the eyes of the beholder. So, what may be a double top to me, may not be to someone else. I tried to maintain some rigor by making sure that I only included extremely obvious patterns. That said, the time frame is still kind of small to draw sizeable conclusions. As well, my results contradict other, peer reviewed research. However, I have actually found some recent ish research that does support technical analysis and chart patterns. So I am not completely going against current research with this conclusion. And these results tend to support my personal experience trading these patterns.
Also, it is important to understand that this analysis was done with SPY and SPY alone. Thus, I can’t say that these patterns will have the same result on other equities like TSLA, it would only be generalizable to SPY/SPX.
However, I can say for certain based on this study that over the past 2 years and 3 and a half months, double top patterns, head and shoulders patterns and triple top patterns were correct 83.3% of the time.
And that's it.
Leave your questions/comments/critiques below!
Thanks for reading!
Most of the literature says that chart patterns and technical analysis are not useful and that chart patterns are the circumstances of randomness. Most research argues that you cannot dictate what a stock will do based on the candlestick charts and patterns, as these only offer retro-spective perspectives in the stock and are not able to be used prospectively.
I, as someone with formal education in statistics and also as a full time trader, am torn between these analyses. On one hand, I can see where the researchers are coming from, on the other hand, I know it can’t be 100% true because many people are successful trading these strategies. While my personal strategy relies primarily on math, I would be lying if I said that I did not let chart patterns influence my decisions to some extent.
Also, the majority of people conducting these studies are not themselves traders. I, as someone who published and familiar with conducting research studies, figured meh. Why not get it straight from the horses mouth and do this myself. So alas, here we go!
I will present this as academically and rigorously as possible!
Procedure:
I looked at ONLY bearish chart patterns, particularly the patterns: head and shoulders, double top and triple tops. This was conducted ONLY using the 1 hour time frame, as the 5 minute time-frame would have been too labour intensive.
Additionally, I had included bear flags in the bearish chart patterns, however bear flags usually accompanied head and shoulders and double tops, which were more visible first and thus, I ended up not having any cases of bear flags to include.
The time frame was 1 hour and I included just over 2 years of data, from 2020 through till Friday.
The data was collected using TradingView’s platform and analyzed using SPSS v25.
Definition of Bearish Chart Pattern
I was somewhat lax with my interpretation of double tops and head and shoulders; however, the major inclusion criteria is that they had to be clearly interpretable as the bearish chart pattern. If I had to look and analyze it in-depth with lines, etc., then it was not included. It did not need to be perfect, it could be lopsided, it just needed to be easily recognizable as the pattern. However, Heikin Ashi candles were used to easily identify the patterns and verified via line charts.
Results
Between January 2020 and April 2022, on the 1 hour time frame, there were a total of 24 clearly observable bearish chart patterns. The breakdown is as follows:
Double Top: 12 incidents (50%)
Head and Shoulders: 9 incidents (37.5%)
Triple Top: 3 incidents (12.5%).
The success rate of all patterns was 83.3%.
The most successful chart pattern was a triple top. Of the three documented incidents, the success rate was 100%. The next was head and shoulders with a success rate of 89% (8 passes and 1 fail). The least successful pattern was the double top with a success rate of 75% (9 successes and 3 fails).
The results, and this is the KEY point, WERE in fact statistically significant via analysis through T-Test with a P value of 0.001!!!!! This means that these results cannot be attributable to chance or randomness and in fact, have statistically significant meaning.
Out of all the patterns, the average sell off that accompanied it was a drop of 10 points (standard deviation 11.27). Of the instances that failed, the average rise in price was 9.72 points (standard deviation of 4.66).
Below is a list of the average point sell off by chart pattern when successful:
Head and Shoulders: -12.33 (SD = 9.58)
Double Top: -8.59 (SD = 14.03)
Triple Top: -9.06 (SD = 5.56)
Conclusion / Limitations:
Please note, that this is far from what I would deem scholarly research. The problem with this type of research is trading is highly subjective and in the eyes of the beholder. So, what may be a double top to me, may not be to someone else. I tried to maintain some rigor by making sure that I only included extremely obvious patterns. That said, the time frame is still kind of small to draw sizeable conclusions. As well, my results contradict other, peer reviewed research. However, I have actually found some recent ish research that does support technical analysis and chart patterns. So I am not completely going against current research with this conclusion. And these results tend to support my personal experience trading these patterns.
Also, it is important to understand that this analysis was done with SPY and SPY alone. Thus, I can’t say that these patterns will have the same result on other equities like TSLA, it would only be generalizable to SPY/SPX.
However, I can say for certain based on this study that over the past 2 years and 3 and a half months, double top patterns, head and shoulders patterns and triple top patterns were correct 83.3% of the time.
And that's it.
Leave your questions/comments/critiques below!
Thanks for reading!
For real-time updates and premium indicators, consider joining my group at: patreon.com/steversteves
Related publications
Disclaimer
The information and publications are not meant to be, and do not constitute, financial, investment, trading, or other types of advice or recommendations supplied or endorsed by TradingView. Read more in the Terms of Use.