I tried the code below, and I got another error : "The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()"

        high1 = history(15, "1d", "high")[:-1]  
        low1 = history(15, "1d", "low")[:-1]  
        close= history(15, "1d", "price")[:-1]  
        adx = talib.ADX(high1[s], low1[s], close[s] , timeperiod=14)

         if adx > 30:  

Hi Youness,

Here's a version of talib.ADX that works. Unfortunately, talib functions don't work well in pipeline because you would have to iterate through each security in a for loop in a custom factor calling the talib function. This would be very inefficient and lead to a very slow backtest.

I hope this helps.

Cheers,
Jamie

Thanks Jamie.

I'm using "set_universe"and I'm getting this error: "ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()"

import talib

def initialize(context):  
    set_universe(universe.DollarVolumeUniverse(90, 91))  
    #context.size = float(context.portfolio.cash / 30.0)  
    #context.shorts =  context.size * 0.5  
    context.shorting= True  
    context.max_positions = 10  
    #context.max_positions2 = 6

    #schedule_function(logic,date_rules.every_day(),time_rules.market_close())  

def handle_data(context, data):  
    context.l = float(( 0.5 * context.portfolio.portfolio_value) /7)  
    #context.s =  float(( -0.5 * context.portfolio.portfolio_value) /7)  
    open_orders = get_open_orders()  
    position_count = get_position_count(context)  
    for s in data:  
        if s in open_orders:  
            continue  
        #price = data[s].price  
        current_position = context.portfolio.positions[s].amount  
        price = data[s].price  
        high1 = history(15, "1d", "high")  
        low1 = history(15, "1d", "low")  
        close= history(15, "1d", "price")  
        #highs=high1[s][-1]  
        #lows=low1[s][-1]  
        #closes=close[s][-1] 

        adx = talib.ADX(high1[s], low1[s], close[s] , timeperiod=14)  
        if position_count < context.max_positions:  
            if  (current_position == 0)    :  
                        if (adx > 35) :  
                            order_target_value(s, context.l )  
                            log.info("LONG " + str(s.symbol))  
                            position_count +=1  

    record(Leverage = context.account.leverage)  
def get_position_count(context):  
    n = 0  
    for position in context.portfolio.positions.itervalues():  
        if position.amount != 0:  
            n += 1  
    return n

Hi Youness,

This error is being raised because you are comparing adx which is an array of rolling values to an int (35). Check out the documentation on talib. You probably want adx[-1] in that line!

Thanks again Jamie.

I've added [-1], it runs for a while then " Exception: inputs are all NaN ".

        high1 = history(15, "1d", "high")  
        low1 = history(15, "1d", "low")  
        close= history(15, "1d", "price")  
        adx = talib.ADX(high1[s], low1[s], close[s] , timeperiod=14)  
        if position_count < context.max_positions:  
            if  (current_position == 0)    :  
                        if (adx[-1]  > 35) :  

Hi Youness,

The talib.ADX function calculates rolling ADX values, and timeperiod corresponds to the rolling window length. It seems your timeperiod is too long for the length of data you're providing. Try increasing your history lookback, or decreaseing the timeperiod!

Hi Jamie.

I've fixed it by adding:

        high1= high1.fillna(0)  
        low1= low1.fillna(0)  
        close = close.fillna(0)  

Thanks!