ANALYZING THE ALLEGED REVERSAL PATTERN
IntroductionThere are several tools and technical indicators to analyze price movements in charts. Whether it may be candlestick formations, the relative strength index, moving averages or other approaches, traders are constantly searching for the best method to predict future price movements. However, most of the patterns shown by these indicators are only said to have certain impacts on future price movements without having been statistically proven for cryptocurrencies yet. Therefore, I want to have an in-depth look at a certain candlestick formation in this article and examine, whether its presumed impact can be proven by numbers.
HammerThe candlestick formation “Hammer” is one of the most widely known candlestick patterns. It is a candlestick with a relatively small body that is accompanied by a long wick to the downside and a short to non-existent wick to the upside. The hammer visualizes an interval in a chart, in which lots of selling pushed the price down initially, before the bulls pushed the price back up, thus positioning the closing price relatively near to the opening price.
Two types of the hammer candle allegedly have a predictive impact on prices. One is a hammer candle that is positioned in a chart after a downtrend. It is assumed to indicate the end of a downtrend and is widely recognized as a bullish reversal pattern. The second hammer pattern is called “hanging man”, which is a hammer that appears after an uptrend and is supposed to be a bearish reversal pattern.
While the human eye might spot hammer candles very easily on a chart, I had to conduct a mathematical definition for the pattern in this analysis. This definition is determined by these three points:
- The wick to the upside has to be smaller than 25% of the size of the candle body.
- The wick to the downside has to be bigger than 200% of the size of the candle body.
- The range from highest price to lowest price of the candle has to be bigger than 75% of the average range from highest to lowest price of the last 7 candles before in this chart. This rule was set in place to exclude very small hammer candles.
Data SampleAfter defining the candle formation, we have to collect data to examine, whether the hammer candle acts as a reversal pattern in crypto. Therefore, I collected daily trading data of 19 major cryptocurrencies against USD (BTC, ETH, XRP, BCH, LTC, EOS, BNB, XMR, TRX, XLM, ADA, XTZ, LINK, NEO, ETC, DASH, IOTA, DOGE and BAT). I restricted the data sample to these 19 cryptos, since I only wanted to include assets that had decent trading volumes for a longer period and were listed on multiple exchanges. The result is a data sample with 27,729 records.
FindingsOf these 27,729 records, 402 fit the definition of a hammer/hanging man candle. That is around 1.45%, meaning we saw a hammer/hanging man candle on average every 69 days in a major crypto chart against USD. Comparing the closing price of the candle to daily closing prices after it occurred, we see an average increase of 0.1% after 1 day, 0.52% after 3 days and 2.13% after 7 days (Median: -0.54% after 1 day, -0.69% after 3 days, -0.89% after 7 days). Not surprisingly, the percentage values are not very conclusive, because they contain both a presumed bullish reversal and a presumed bearish reversal.
If we now want to separate hammer and hanging man candle from each other, it gets a little more complicated. A hammer candle only occurs, if there was a downtrend before the candle, while a hanging man only occurs, if there was an uptrend in the days prior to the candle. To define an uptrend/downtrend prior to the candle, we have to define a fixed timespan, in which the trend appeared. We also have to define a minimum value for the size of the trend to exclude tiny movements.
Accordingly, I started with 1-day movements. A hammer candle is identified, if the opening price of the day prior was higher than the hammer candle’s closing price by at least the size of the hammer’s candle body. Subsequently, a hanging man candle is identified, if the opening price of the day prior is lower than the hanging man’s candle by at least the size of the hanging man’s candle body. This 1-day algorithm found 183 hammer candles and 101 hanging man candles. The results are not very promising though: On average, the 1-day hammer candles led to a price increase of 0.47% compared to 1 day later, which barely confirms the bullish reversal pattern. Surprisingly, the 1-day hanging man pattern also led to a price increase, namely 1.35%, compared to the closing price 1 day later. This number shows the exact opposite effect to the bearish reversal that this pattern should have.
As a summary, the 1-day hammer/hanging man candles were not indicative of any trends. Therefore, I tested 3-day and 7-day movements that followed the same calculation than the 1-day movements but with 3-day and 7-day intervals. The algorithm identified 137 3-day hammer candles, 89 3-day hanging man candles, 78 7-day hammer candles and 63 7-day hanging man candles. The results are again different than expected: The 3-day hammer led to an average price decline of 2.03% while the 7-day hammer led to a price decline of 2.63%. The 3-day hanging man led to a price increase of 3.89%, while the 7-day hanging man led to a price increase of 1.9%. In both cases the results contradict the assumed effects.
ConclusionAlthough the hammer/hanging man candle is said to be a trend reversal candlestick pattern, the underlying analysis of this article could not prove a correlation between the appearance of the pattern and a trend reversal. The hammer candles that should have signaled a bullish reversal actually led on average to a further price decline, while the hanging man candles, which should have proven a bearish reversal, actually led on average to a further price increase. Concluding from that, trades that were based on the assumed impact of hammer and hanging man candles did not necessarily turn out profitable in the considered data sample. However, the occurrence of both candle patterns has been rare, so that a bigger data sample could provide different results. Nonetheless, the two candle patterns could not be proven to be good predictive indicators for past cryptocurrency price movements, which in my personal opinion signals that they shouldn’t be given too much recognition in a chart prediction.
Articles, that may interest you: