Error: inputs are all NaN

Back to Community

posted Nov 23, 2017

Good afternoon,

I am a student and I was trying to implement the WaveTrend Oscillator strategy: https://www.tradingview.com/script/2KE8wTuF-Indicator-WaveTrend-Oscillator-WT/

This is my code and I'm kind of stuck at the moment because it keeps giving me error and I'm not able to run a backtest. It seems like it doesn't recognise the local variables wt1 and wt2 even though I use them in the "open_positions" function right after. what I wanted to do is selling AAPL when the indicator is high and buying it when is low. Can anyone help me?

import talib  
# ---------------------------------------------------  
n1, n2, period = 10, 21, 12  
# ---------------------------------------------------  
def initialize(context):  
    context.stock = sid(24)  
    schedule_function(open_positions, date_rules.week_start(), time_rules.market_open())  
def handle_data(context, data):  
    if get_open_orders(): return  
    close = data.history(context.stock, 'close', period + 1, '1d')  
    low = data.history(context.stock, 'low', period + 1, '1d')  
    high = data.history(context.stock, 'high', period + 1, '1d')  
    ap = (high+low+close)/3  
    esa = talib.EMA(ap, timeperiod=n1)  
    d = talib.EMA(abs(ap - esa), timeperiod=n1)  
    ci = (ap - esa) / (0.015 * d)  
    wt1 = talib.EMA(ci, timeperiod=n2)  
    wt2 = talib.SMA(wt1, timeperiod=4)

def open_positions(context, data):  
    if data.can_trade(context.stock <  wt1):  
        order_target_percent(context.stock, 2)  
    elif data.can_trade(context.stock > wt2):  
        order_target_percent(context.stock, -1)

Thanks in advance,
Mattia

2 responses

Luca

Nov 23, 2017

The variables defined in one function are not visible in other functions, but you can save wt1 and wt2 in context and access them in open_positions function. Also I fixed few small things but still esa ad d are mostly nan, you have to double check the logic of your computation.

import talib  
import numpy as np  
# ---------------------------------------------------  
n1, n2, period = 10, 21, 12  
# ---------------------------------------------------  
def initialize(context):  
    context.stock = sid(24)  
    schedule_function(open_positions, date_rules.week_start(), time_rules.market_open())  
def handle_data(context, data):  
    if get_open_orders(): return  
    close = data.history(context.stock, 'close', period + 1, '1d')  
    low = data.history(context.stock, 'low', period + 1, '1d')  
    high = data.history(context.stock, 'high', period + 1, '1d')  
    ap = (high+low+close)/3.  
    esa = talib.EMA(ap, timeperiod=n1)  
    d = talib.EMA(np.abs(ap.values - esa), timeperiod=n1)  
    ci = (ap - esa) / (0.015 * d)  
    wt1 = talib.EMA(ci, timeperiod=n2)  
    wt2 = talib.SMA(wt1, timeperiod=4)  
    context.wt1 = wt1  
    context.wt2 = wt2

def open_positions(context, data):  
    if data.can_trade(context.stock <  context.wt1):  
        order_target_percent(context.stock, 2)  
    elif data.can_trade(context.stock > context.wt2):  
        order_target_percent(context.stock, -1)

Luca

Nov 23, 2017

Consider the line :

esa = talib.EMA(ap, timeperiod=n1)

With n1= 10 len(ap)= 13, you get 9 NaNs:

[ nan nan nan nan nan nan nan nan nan 323.49433333 323.29718182 324.20175482 325.64628425]

With n1=4 you get 3 NaNs

[ nan nan nan 321.70833333 322.961 323.36193333 323.57182667 324.435096 324.90772427 324.57396789 323.70838074 325.53396177 328.17904373]

This is because the period tells the function how many items you need before computing a value, so if n1 is 10, you need at least 10 elements before computing the value, that's the reason why the first 9 elements are NaNs

You've successfully submitted a support ticket.

Our support team will be in touch soon.