Getting RSI in python

Question

I've been trying to calculate the 14 RSI of stocks and I managed to get it to work, somewhat, it gives me inaccurate numbers

import pandas as pd 
import datetime as dt
import pandas_datareader as web

ticker = 'TSLA'

start = dt.datetime(2018, 1, 1)
end = dt.datetime.now()

data = web.DataReader(ticker, 'yahoo', start, end)

delta = data['Adj Close'].diff(1)
delta.dropna(inplace=True)

positive = delta.copy()
negative = delta.copy()

positive[positive < 0] = 0
negative[negative > 0] = 0

days = 14

average_gain = positive.rolling(window=days).mean()
average_loss = abs(negative.rolling(window=days).mean())

relative_strenght = average_gain / average_loss
rsi = 100.0 - (100.0 / (1.0 + relative_strenght))
print(ticker + str(rsi))

It ends up giving me 77.991564 (14 days RSI) when I should be getting 70.13 (14 days RSI), does any know what I'm doing wrong?

also yes I've read Calculating RSI in Python but it doesn't help me with what I need

Do you need to calculate it yourself? there are libraries that do that for you. pandas-ta and ta are couple that are widely used. — Jonathan Leon, Jan 15 '21 at 04:52

score 1 · Answer 1 · answered Jan 28 '21 at 21:45

Here is one way to calculate by yourself RSI. The code could be optimized, but I prefer to make it easy to understand, and the let you optimize.

For the example, we assume that you've got a DataFrame called df, with a column called 'Close', for the close prices. By the way, notice that if you compare results of the RSI with a station, for example, you should be sure that you compare the same values. For example, if in the station, you've got the bid close, and that you calculate by your own on the mid or the ask, it will not be the same result.

Let's see the code :

def rsi(df,_window=14,_plot=0,_start=None,_end=None):
    """[RSI function]

    Args:
        df ([DataFrame]): [DataFrame with a column 'Close' for the close price]
        _window ([int]): [The lookback window.](default : {14})
        _plot ([int]): [1 if you want to see the plot](default : {0})
        _start ([Date]):[if _plot=1, start of plot](default : {None})
        _end ([Date]):[if _plot=1, end of plot](default : {None})
    """    

    ##### Diff for the différences between last close and now
    df['Diff'] = df['Close'].transform(lambda x: x.diff())
    ##### In 'Up', just keep the positive values
    df['Up'] = df['Diff']
    df.loc[(df['Up']<0), 'Up'] = 0
    ##### Diff for the différences between last close and now
    df['Down'] = df['Diff']
    ##### In 'Down', just keep the negative values
    df.loc[(df['Down']>0), 'Down'] = 0 
    df['Down'] = abs(df['Down'])

    ##### Moving average on Up & Down
    df['avg_up'+str(_window)] = df['Up'].transform(lambda x: x.rolling(window=_window).mean())
    df['avg_down'+str(_window)] = df['Down'].transform(lambda x: x.rolling(window=_window).mean())

    ##### RS is the ratio of the means of Up & Down
    df['RS_'+str(_window)] = df['avg_up'+str(_window)] / df['avg_down'+str(_window)]

    ##### RSI Calculation
    ##### 100 - (100/(1 + RS))
    df['RSI_'+str(_window)] = 100 - (100/(1+df['RS_'+str(_fast)]))

    ##### Drop useless columns
    df = df.drop(['Diff','Up','Down','avg_up'+str(_window),'avg_down'+str(_window),'RS_'+str(_window)],axis=1)

    ##### If asked, plot it!
    if _plot == 1:
        sns.set()
        fig = plt.figure(facecolor = 'white', figsize = (30,5))
        ax0 = plt.subplot2grid((6,4), (1,0), rowspan=4, colspan=4)
        ax0.plot(df[(df.index<=end)&(df.index>=start)&(df.Symbol==_ticker.replace('/',''))]['Close'])
        ax0.set_facecolor('ghostwhite')
        ax0.legend(['Close'],ncol=3, loc = 'upper left', fontsize = 15)
        plt.title(_ticker+" Close from "+str(start)+' to '+str(end), fontsize = 20)

        ax1 = plt.subplot2grid((6,4), (5,0), rowspan=1, colspan=4, sharex = ax0)
        ax1.plot(df[(df.index<=end)&(df.index>=start)&(df.Symbol==_ticker.replace('/',''))]['RSI_'+str(_window)], color = 'blue')
        ax1.legend(['RSI_'+str(_window)],ncol=3, loc = 'upper left', fontsize = 12)
        ax1.set_facecolor('silver')
        plt.subplots_adjust(left=.09, bottom=.09, right=1, top=.95, wspace=.20, hspace=0)
        plt.show()
    return(df)

To call the function, you just have to type

df = rsi(df)

if you keep it with default values, or to change _window and/or _plot for the arg. Notice that if you input _plot=1, you'll need to feed starting and ending of the plot, with a string or a date time.

Getting RSI in python

1 Answers1