origin
When the market plummeted on July 26 this year, a big guy in the Hulu Group came up with a picture. I was very shocked when I read it because the author predicted this decline:
Although it is not always correct, the big guys in the group said it was a magical book. Out of curiosity, I opened it and read it, and found that the author concluded that "A-shares have a 10-year cycle." Even though they did not understand the author's argument, But at a glance it seems to make sense, so let’s verify it!
Author’s opinion and verification
Author's point of view
The author's point of view: China's stock market follows a 10-year cycle. A bull market begins in Year C of the lunar calendar . Year D continues to Year Xin after adjustment. Year Rengui is a bear market, and Year B breeds a bull market.
Since the author's calculation method is based on the Heavenly Stems calendar, and there are 10 Heavenly Stems, each year can correspond to the mantissa of the AD calendar year, which can be organized into this table:
Time flies | The last digit of AD | market conditions |
---|---|---|
Geng | 0 | structure bear |
Spicy | 1 | bear market |
the ninth of the ten Heavenly Stems | 2 | bear market |
Gui | 3 | bear market |
First | 4 | Bear to Bull conversion |
Second | 5 | Bear to Bull conversion |
C | 6 | ox |
Man | 7 | Bull+Bear conversion |
E | 8 | big bear |
Has | 9 | ox |
verify
The content of the preliminary verification is the ten-year cycle proposed by the author . It is not the previous content that is accurate to the month and day. The main reason is that the first impression given by those things cannot be 100% correct, and if this cycle has a certain general If there are rules to follow, it will be of great help to the timing of our strategies.
My verification idea is:
- Download the data of the Shanghai Composite Index and the CSI 300 Index (this is covered in the course, if you are too lazy to crawl, you can go to the website) and convert it into the chronology of the zodiac signs and branches;
- Calculate the rise and fall, amplitude, number of rising days, number of falling days and other indicators for each year (classified by year of stems and branches, the same below), and make a preliminary judgment based on the author's conclusion;
- Draw annual charts and make subjective judgments (didn’t we agree that we are doing quantification);
Preparation
- Crawl index data
At present, the earliest data we can obtain for the Shanghai Composite Index is December 1990 (from Sina Finance), and the earliest data for the CSI 300 Index is April 2005.
- Download a library that can convert to lunar time
I use this library
The converted CSI 300 Index looks like this:
process
First, let’s make descriptive statistics based on the chronology of the stems and branches:
df_all = pd.DataFrame()
df_all['year'] = list(set(df001['year']))
for i in list(set(df001['year'])):
df_all.loc[df_all['year'] == i, 'candle_begin_time'] = df001.loc[df001['year'] == i, 'candle_end_time'].iloc[0]
df_all.loc[df_all['year'] == i, 'candle_end_time'] = df001.loc[df001['year'] == i, 'candle_end_time'].iloc[-1]
df_all.loc[df_all['year'] == i, '涨跌幅'] = round(df001.loc[df001['year'] == i, 'close'].iloc[-1] / df001.loc[df001['year'] == i, 'close'].iloc[0] - 1, 2)
df_all.loc[df_all['year'] == i, '振幅'] = round(df001.loc[df001['year'] == i, 'close'].max() / df001.loc[df001['year'] == i, 'close'].min(), 2)
df_all.loc[df_all['year'] == i, '上涨天数'] = len(df001.loc[(df001['year'] == i) & (df001['close_change'] >= 0), 'close_change'] )
df_all.loc[df_all['year'] == i, '下跌天数'] = len(df001.loc[(df001['year'] == i) & (df001['close_change'] < 0), 'close_change'] )
df_all['上涨天数-下跌天数'] = df_all['上涨天数'] - df_all['下跌天数']
df_all.sort_values(['candle_begin_time'])
The result is as follows:
year | candle_begin_time | candle_end_time | Quote change | amplitude | rising days | Down days | Number of rising days - number of falling days |
---|---|---|---|---|---|---|---|
Gengwu | 1990/12/19 | 1991/2/14 | 0.07 | 1.07 | 39 | 1 | 38 |
Xin Wei | 1991/2/19 | 1992/1/31 | 0.12 | 1.12 | 180 | 57 | 123 |
Renshen | 1992/2/7 | 1993/1/22 | 7.09 | 10.43 | 69 | 14 | 55 |
Guiyou | 1993/1/27 | 1994/2/4 | -0.31 | 2.04 | 129 | 138 | -9 |
Jiaxu | 1994/2/14 | 1995/1/27 | -0.28 | 3.09 | 100 | 146 | -46 |
Yi Hai | 1995/2/6 | 1996/2/16 | 0.04 | 1.74 | 128 | 138 | -10 |
Bingzi | 1996/3/4 | 1997/1/31 | 0.6 | 2.25 | 129 | 106 | 23 |
Ding Chou | 1997/2/17 | 1998/1/23 | 0.24 | 1.68 | 137 | 99 | 38 |
Wuyin | 1998/2/9 | 1999/2/9 | -0.13 | 1.33 | 118 | 140 | -22 |
Ji Mao | 1999/3/1 | 2000/1/28 | 0.4 | 1.64 | 120 | 111 | 9 |
Gengchen | 2000/2/14 | 2001/1/19 | 0.23 | 1.33 | 136 | 98 | 38 |
Xin Si | 2001/2/5 | 2002/2/8 | -0.25 | 1.65 | 123 | 129 | -6 |
Renwu | 2002/2/25 | 2003/1/29 | -0.02 | 1.31 | 116 | 115 | 1 |
Guiwei | 2003/2/10 | 2004/1/16 | 0.08 | 1.24 | 114 | 118 | -4 |
Jiashen | 2004/1/29 | 2005/2/4 | -0.22 | 1.5 | 113 | 143 | -30 |
Yiyou | 2005/2/16 | 2006/1/25 | -0.02 | 1.3 | 118 | 116 | 2 |
Bingxu | 2006/2/6 | 2007/2/16 | 1.33 | 2.41 | 166 | 91 | 75 |
Dinghai | 2007/2/26 | 2008/2/5 | 0.51 | 2.2 | 150 | 85 | 65 |
Wuzi | 2008/2/13 | 2009/1/23 | -0.56 | 2.73 | 101 | 135 | -34 |
Ji Chou | 2009/2/2 | 2010/2/12 | 0.5 | 1.73 | 157 | 102 | 55 |
Gengyin | 2010/2/22 | 2011/2/1 | -0.07 | 1.34 | 119 | 114 | 5 |
Xin Mao | 2011/2/9 | 2012/1/20 | -0.16 | 1.42 | 112 | 124 | -12 |
Renchen | 2012/1/30 | 2013/2/8 | 0.06 | 1.26 | 133 | 123 | 10 |
Guisi | 2013/2/18 | 2014/1/30 | -0.16 | 1.24 | 110 | 123 | -13 |
Jiawu | 2014/2/7 | 2015/2/17 | 0.59 | 1.7 | 148 | 108 | 40 |
Yiwei | 2015/2/25 | 2016/2/5 | -0.14 | 1.95 | 133 | 104 | 29 |
Bingshen | 2016/2/15 | 2017/1/26 | 0.15 | 1.22 | 130 | 107 | 23 |
Ding You | 2017/2/3 | 2018/2/14 | 0.02 | 1.17 | 154 | 104 | 50 |
Wuxu | 2018/2/22 | 2019/2/1 | -0.2 | 1.35 | 105 | 129 | -24 |
Jihai | 2019/2/11 | 2020/1/23 | 0.12 | 1.23 | 125 | 112 | 13 |
Gengzi | 2020/2/3 | 2021/2/10 | 0.33 | 1.37 | 140 | 115 | 25 |
Xin Chou | 2021/2/18 | 2021/7/30 | -0.08 | 1.1 | 54 | 58 | -4 |
The following is the situation of CSI 300:
year | candle_begin_time | candle_end_time | CSI 300 rise and fall | CSI 300 Amplitude | CSI 300 rising days | CSI 300 falling days | CSI 300 rising days - falling days |
---|---|---|---|---|---|---|---|
Yiyou | 2005/2/16 | 2006/1/25 | 0.01 | 1.23 | 103 | 93 | 10 |
Bingxu | 2006/2/6 | 2007/2/16 | 1.59 | 2.67 | 166 | 91 | 75 |
Dinghai | 2007/2/26 | 2008/2/5 | 0.82 | 2.39 | 151 | 84 | 67 |
Wuzi | 2008/2/13 | 2009/1/23 | -0.58 | 3.08 | 104 | 132 | -28 |
Ji Chou | 2009/2/2 | 2010/2/12 | 0.58 | 1.84 | 164 | 95 | 69 |
Gengyin | 2010/2/22 | 2011/2/1 | -0.05 | 1.41 | 119 | 114 | 5 |
Xin Mao | 2011/2/9 | 2012/1/20 | -0.18 | 1.48 | 109 | 127 | -18 |
Renchen | 2012/1/30 | 2013/2/8 | 0.13 | 1.32 | 130 | 126 | 4 |
Guisi | 2013/2/18 | 2014/1/30 | -0.2 | 1.27 | 105 | 128 | -23 |
Jiawu | 2014/2/7 | 2015/2/17 | 0.59 | 1.75 | 140 | 116 | 24 |
Yiwei | 2015/2/25 | 2016/2/5 | -0.15 | 1.88 | 132 | 105 | 27 |
Bingshen | 2016/2/15 | 2017/1/26 | 0.15 | 1.24 | 126 | 111 | 15 |
Ding You | 2017/2/3 | 2018/2/14 | 0.18 | 1.32 | 148 | 110 | 38 |
Wuxu | 2018/2/22 | 2019/2/1 | -0.2 | 1.39 | 105 | 129 | -24 |
Jihai | 2019/2/11 | 2020/1/23 | 0.21 | 1.27 | 121 | 116 | 5 |
Gengzi | 2020/2/3 | 2021/2/10 | 0.57 | 1.65 | 147 | 108 | 39 |
Xin Chou | 2021/2/18 | 2021/7/30 | -0.17 | 1.22 | 59 | 53 | 6 |
Then we took a rough look at it based on the table above. I browsed it again and found that it was basically the same.
For example, the author emphasized that the year of Bing and Ding was the starting point of a bull market. In 1996 and 2006, there was an increase of more than 50%. In 1997 and 2007, the index also increased by more than 20%. Only in 2016 and 2017, the increase was smaller. But they are all positive; then look at the correction growth rates after the bull market in Year 5 are all negative; and in Year 9, the bull market specified by the author, there are also relatively large increases, but relatively speaking, the increase in 2019 is smaller; Then look at the Geng Year, it is said that there was a market, and only one of the four times was negative; only the Renshen Year of 1992 was originally a bear market, but the market performed very well, and the final increase was 7 times (but this year is quite special, specifically I don’t know why).
Ok, after the first round of judgment, we tentatively think that the author's conclusion is reasonable to a certain extent. Next, we will further understand it in the form of charts.
plt.figure(figsize=(15, 38))
var1 = ['丙', '丁', '戊', '己', '庚', '辛', '壬', '癸', '甲', '乙']
var_temp = 0
for o in var1:
ax = plt.subplot(10, 1, var_temp+1)
for i in list(filter(lambda x: x.startswith(o), set(df001['year']))):
temp = df001.loc[df001['year']==i, 'close'].reset_index(drop=True)
temp['curve'] = (1+temp.pct_change()).cumprod()
ax.plot(temp['curve'], label=i)
plt.legend()
var_temp += 1
plt.savefig('picture.png')
plt.show()
Get a picture like this:
This is the CSI 300:
At this point, I found that if I want to affirm the author's conclusion, some years obviously do not support it. For example, the Renshen year soared in 1992, and the author proposed that the Renshen year was a bear market. If I want to deny the author's conclusion, there is insufficient evidence. , the author’s conclusion that years Bing and D were all bull markets, Years Wu was a bear market, and Year Ji was a bull market are indeed the same. It’s just that the so-called bull markets in recent years have seen relatively small increases and decreases, so there are also explanations for the different geological zones. The combinations come in different strengths and weaknesses. For example, the author mentioned that the year 26 is the Bingwu year, the noon fire is the purest fire, and the heavenly stems and earthly branches are all wealth, so it is a very strong bull market.
in conclusion
As for what it does for our strategy? I personally haven’t found a particularly good auxiliary method. The only useful method is to avoid the big drop after the big rise in Ding year, that is, Wu year. But in the bear market in the third year of Xin Rengui mentioned by the author, I think many strategies can make money. , and the big bull market in years B and D was not the time when each strategy performed best. The author said that the year A and B during the bear-bull transition were the years when many strategies made the most money.
In addition, not every year B and D is a bull market. For example, the recent one is not. It is a bull market composed of years A and B in 14 and 15. 16 and 17 are just small bulls that went from 2,600 to 3,500.
In short, the current sample data is too small, only three cycles, and the author's description of the cycle is not very quantitative. Our verification cannot 100% confirm or deny his conclusion. There is uncertainty in using his policy completely. Of course, everyone has different opinions on benevolence and wisdom.