确定支撑/阻力水平和趋势线

Question

I don't need a code just the logic on how to do this. Example of S/R levels and a trendline

I know s/r levels differ from one person to another so I have attached a graph to show what criteria I use to consider a support/resistance level or a trend line.

• S/R level has to have more than 2 touches if there are more than over fits this criterion then the one with most touchdowns is the valid one, same for resistance.

• A trend line has to have 3 touches in order to be valid.

All lines have a margin error since these lines can get extended. and of course, once a support line is broken it turns into resistance.

I tried taking an array of min values then see if other bars fall within a certain range of these values but it wasn't accurate enough!

Answer 1

Below is something for you to start with. Emphasis on the 'start' because it doesn't directly answer your questions.

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
%matplotlib inline

Simulate data.

np.random.seed(42)
mu = 0.03
ret = mu+np.random.randn(500)
price = 1+np.cumsum(ret)

Look for support by first, removing noise from signal with moving average. Offset by support value, then look for locations where offset value is at zero, which would be the locations touching the support line .

def get_support_line(df,days=100):
    df['price_ma']=df['price'].rolling(5).mean()
    df = df.dropna()
    supports={}

    for x in np.arange(0,len(df)-days,10):
        price_tmp = df.iloc[x:x+days]['price_ma'].values
        # offset signal by support value, loc crossing 0 will be support.
        support_val = price_tmp[-1]
        norm = price_tmp-support_val
        lim = 0.05 # hack to find intersect with support line with tolerance
        signal = np.logical_and(norm<lim,norm>-1*lim).astype(np.int)
        inds = np.where(np.diff(signal)>0)[0]
        if np.sum(signal)>3:
            supports[x+days]={
                'x':inds+x,
                'y':support_val*np.ones(inds.shape)
            }
    return df, supports

df = pd.DataFrame()
df['price']=price
df, supports = get_support_line(df)

Plot to verify logic. You now should start tweaking the free parameters and also the logic to your "liking".

df.plot(figsize=(10,5))
for k,v in supports.items():
    print('support identified at days prior index',k,'value: {:1.2f}'.format(v['y'][0]),'touch inds:',v['x'])
    plt.plot([v['x'][0],v['x'][-1]],[v['y'][0],v['y'][-1]],color='red')
plt.grid(True)

I don't buy technical analysis, but I think this is an interesting problem, implementation wise! There should be lots of libraries out there. so please be sure to do a google search, and come back and share your solutions and findings.