Python TA-Lib RSI wrong results

Question

I'm trying to get the RSI of a stock using TA-Lib in python and it keeps giving me wrong numbers.

import talib
import pandas as pd
from td.client import TDClient

ticker = 'GOOG'
data = TDSession.get_price_history(
    symbol = ticker,
    period_type = 'month',
    frequency_type = 'daily',
    frequency = 1,
    period = 1,
)

df = pd.DataFrame(data['candles'])

close = df['close']

# Gets the RSI of the ticker 
rsi = str(talib.RSI(close, timeperiod=14))

current = float(rsi[len(rsi)-40:len(rsi)-33])

Is there a fix for this?

user_11534493 · Answer

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.

You should then compare it to Ta-Lib.

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):
    """[RS functionI]

    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.

Python TA-Lib RSI wrong results

1 Answers