Crossover error operator '-'

Xavier Escudero

Using btalib crossover an error is shown:

Library/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/pandas/core/computation/expressions.py:200: UserWarning: evaluating in Python space because the '-' operator is not supported by numexpr/Library/Frameworks/Python.framework/Versions/3.8/lib/python3.8/site-packages/pandas/core/computation/expressions.py:200: UserWarning: evaluating in Python space because the '-' operator is not supported by numexpr for the bool dtype, use '^' instead

My code uses as input parameter a pandas dataframe:

if len(df)>50:
     sma50 = btalib.sma(df, period=50)
     df['sma50'] = btalib.sma(df, period=50).df

     if len(df)>200:
           sma200 = btalib.sma(df, period=200)
           df['sma200'] = sma200.df      
           btalib.crossover(sma50, sma200)

run-out

Could you share a few rows of your dataframe please?

Xavier Escudero

@run-out Here you got it (last columns are sma20 and sma50):

o h l c v d s symbol max52w min52w sma20 sma50
ts
2020-02-18 85.110069 85.527370 83.987318 84.245651 2872500 0.0 0 A NaN NaN NaN NaN
2020-02-19 84.156212 85.288892 82.914237 84.802040 4741300 0.0 0 A NaN NaN NaN NaN
2020-02-20 84.354941 84.444359 82.824826 83.798531 2543600 0.0 0 A NaN NaN NaN NaN
2020-02-21 83.361355 84.543710 82.864565 84.523842 1762500 0.0 0 A NaN NaN NaN NaN
2020-02-24 81.692144 82.059763 79.506263 79.983185 2919200 0.0 0 A NaN NaN NaN NaN
... ... ... ... ... ... ... .. ... ... ... ... ...
2020-11-24 118.959999 118.959999 111.430000 114.680000 5559900 0.0 0 A NaN NaN 108.683499 104.917219
2020-11-25 116.089996 116.959999 113.410004 114.349998 2862500 0.0 0 A NaN NaN 109.317999 105.209415
2020-11-27 113.769997 114.980003 112.940002 114.089996 983700 0.0 0 A NaN NaN 109.921499 105.502801
2020-11-30 113.889999 117.070000 113.669998 116.900002 3778400 0.0 0 A NaN NaN 110.661999 105.856778
2020-12-01 117.580002 118.379997 115.040001 115.360001 1753700 0.0 0 A NaN NaN 111.167999 106.204910

run-out

@Xavier-Escudero said in Crossover error operator '-':

btalib.crossover(sma50, sma200)

I took a stab at this and I'm getting the same error. I modified the code slightly:

df = pd.read_csv("/home/runout/projects/data/es/es18_20.csv")
df.columns = ['Date', 'High', 'Low', 'Open', 'Close', 'Cumvol', 'Volume']
df = df.set_index('Date')
df.index = pd.to_datetime(df.index)
sma50 = btalib.sma(df, period=50)
df['sma50'] = sma50.df

sma200 = btalib.sma(df, period=200)
df['sma200'] = btalib.sma(df, period=200).df
df = df.dropna()
df
btalib.crossover(sma50, sma200).df

It works right up to the crossover point then errors. I get the same error if I just try using open and close to cross over.

TypeError: numpy boolean subtract, the `-` operator, is not supported, use the bitwise_xor, the `^` operator, or the logical_xor function instead.

So I investigated this and it is a known problem. Here is an example of someone fixing this.

 WillPastor commented on Apr 23, 2019

@zockerwurf @NathanielCustom
I fixed this problem by changing a single character in dejavu/fingerprint.py and it worked fine on my Ubuntu VirtualBox VM with about 8 GB memory. The code that needs to be changed is on line 104 of fingerprint.py ...

    # Boolean mask of arr2D with True at peaks
    detected_peaks = local_max - eroded_background

Needs to be changed to:

    # Boolean mask of arr2D with True at peaks
    detected_peaks = local_max ^ eroded_background

So I found the likely culprit in the btalib file btalib/indicators/crossover.py in the crossover class:

class crossover(_crossbase):
    '''
    This indicator gives a signal if the provided datas (2) cross up or down.

      - `1.0` if the 1st data crosses the 2nd data upwards
      - `-1.0` if the 1st data crosses the 2nd data downwards

    It does need to look into the current time index (0) and the previous time
    index (-1) of both the 1st and 2nd data

    Formula:
      - crossup = data0(-1) < data1(-1) and data0 > data1
      - crossdown = data0(-1) > data1(-1) and data0 < data1
      - crossover = crossup - crossdown
    '''
    group = 'utils'
    alias = 'CrossOver'
    outputs = 'crossover'

    _updown = 0

    def __init__(self):
        self.o.crossover = self._cup - self._cdown

The last line needs to change the - to ^.

    def __init__(self):
        self.o.crossover = self._cup ^ self._cdown

This seems to solve the problem but I have not bug tested it, and this need to be changed in the codebase me thinks.

In my final working version I ended up just using this:

btalib.crossover(sma50.df, sma200.df).df

run-out

I should have checked the pull request for bta-lib, it's a known and ignore issue.

https://github.com/mementum/bta-lib/pull/6/files

Xavier Escudero

@run-out Yes, now that you mention it I remember I saw the pull request, but I don't understand why it's not merged yet (it was commited on june).

Xavier Escudero

@run-out Anyway, thanks for your support, it works!