I might be wrong here, but i think the calculation of the EMA is not entirely correct.

Current code:

def ema(series, periods):
    sma = series.rolling(window=periods, min_periods=periods).mean()[:periods]
    rest = series[periods:]
    return pd.concat([sma, rest]).ewm(span=periods, adjust=False).mean()

Analysis:
Assuming that periods=5. (for easier explanation)
The first line of code results in the first 4 rows to be NaN
and the 5th row to be the SMA of the first 5 rows.
Then you fuse that together with the rest of the series.
This causes the 6th row to be the result of the EMA of:
[None, None, None, None, SMA for first 5 rows, EMA of(4xNone+SMA for first 5 rows)]

Problems:
The 5th row is an SMA.
The following rows are still not a calculation of the expected 5 prices,
but instead of NaN, the SMA and then the actual prices.

I made a quick Jupyter Notebook for a better explanation:
https://colab.research.google.com/drive/1aPb0qPoeNWVLIk_ErhvoZTk1ImNzznCO

My proposition:

def ema(series, periods, fillna=False):
    if fillna is True:
        return series.ewm(span=periods, min_periods=0, adjust=False).mean()

    return series.ewm(span=periods, min_periods=periods, adjust=False).mean()

Hi,
Toady, i ran adx again by fresh install of your ta ,
looks like recently you have performed changes ta lib and adx functionality is not working.

df['adx_pos']=trend.adx_pos( df['high'], df['low'], df['close'], n=8    , fillna=True)
df['adx_neg']=trend.adx_neg(df['high'], df['low'], df['close'], n=8, fillna=True)
df['adx']=trend.adx(df['high'], df['low'], df['close'], n=8,fillna=True)

could you please resolve this issue ASAP. looks like some return get_min_max function not proper.

        adx_pos  adx_neg   adx    high     low    SYMBOL   close   macd  macd_signal    rsi   3_ema  14_ema

date
2019-04-05 0.0 0.0 0.0 768.75 756.10 AXISBANK 762.20 0.00 0.00 50.00 762.20 762.20
2019-04-08 0.0 0.0 0.0 768.00 748.55 AXISBANK 754.60 -0.17 -0.09 0.00 757.64 758.13
2019-04-09 0.0 0.0 0.0 766.00 751.00 AXISBANK 763.40 0.06 -0.03 57.19 760.37 760.14
2019-04-10 0.0 0.0 0.0 767.90 758.00 AXISBANK 760.85 0.07 0.00 48.01 760.57 760.36
2019-04-11 0.0 0.0 0.0 761.35 747.00 AXISBANK 752.30 -0.27 -0.08 29.62 757.39 758.26
2019-04-12 0.0 0.0 0.0 766.75 751.60 AXISBANK 764.35 0.05 -0.05 56.63 759.94 759.67
2019-04-15 0.0 0.0 0.0 770.20 760.00 AXISBANK 762.85 0.18 0.01 53.68 760.97 760.34
2019-04-16 0.0 0.0 0.0 776.00 765.10 AXISBANK 771.10 0.66 0.17 65.21 764.48 762.44
2019-04-18 0.0 0.0 0.0 779.00 769.00 AXISBANK 771.20 0.99 0.36 65.33 766.78 764.05
2019-04-22 20.0 20.0 0.0 768.90 750.55 AXISBANK 755.50 0.40 0.37 40.11 762.95 762.56
2019-04-23 20.0 20.0 0.0 759.30 749.00 AXISBANK 753.25 -0.15 0.25 37.70 759.68 760.99
2019-04-24 20.0 20.0 0.0 755.70 738.35 AXISBANK 752.85 -0.58 0.07 37.25 757.39 759.67
2019-04-25 20.0 20.0 0.0 758.55 738.55 AXISBANK 740.90 -1.56 -0.27 26.25 751.86 756.70
2019-04-26 20.0 20.0 0.0 765.15 747.25 AXISBANK 759.90 -1.18 -0.46 52.16 754.55 757.20
2019-04-30 20.0 20.0 0.0 769.70 755.80 AXISBANK 766.85 -0.46 -0.46 58.34 758.66 758.65
2019-05-02 20.0 20.0 20.0 767.80 748.05 AXISBANK 752.35 -0.79 -0.53 44.50 756.55 757.72
2019-05-03 20.0 20.0 20.0 763.80 753.45 AXISBANK 757.30 -0.72 -0.57 49.25 756.80 757.66
2019-05-06 20.0 20.0 20.0 753.40 744.10 AXISBANK 747.45 -1.28 -0.71 41.17 753.68 756.18
2019-05-07 20.0 20.0 20.0 755.50 740.00 AXISBANK 741.85 -2.06 -0.98 37.17 749.74 754.14
2019-05-08 20.0 20.0 20.0 744.75 733.35 AXISBANK 735.50 -3.04 -1.40 32.98 744.99 751.50
2019-05-09 20.0 20.0 20.0 740.50 729.40 AXISBANK 731.00 -4.07 -1.94 30.20 740.33 748.63
2019-05-10 20.0 20.0 20.0 738.50 729.50 AXISBANK 735.05 -4.54 -2.46 35.82 738.57 746.74
2019-05-13 20.0 20.0 20.0 742.85 730.00 AXISBANK 732.10 -5.05 -2.98 33.55 736.41 744.71
2019-05-14 20.0 20.0 20.0 738.50 722.50 AXISBANK 731.35 -5.44 -3.48 32.94 734.72 742.87
2019-05-15 20.0 20.0 20.0 737.70 716.10 AXISBANK 721.95 -6.34 -4.05 26.05 730.47 740.00
2019-05-16 20.0 20.0 20.0 735.00 718.70 AXISBANK 731.85 -6.26 -4.50 41.03 730.93 738.89
2019-05-17 20.0 20.0 20.0 752.15 727.20 AXISBANK 748.95 -4.92 -4.58 57.99 736.93 740.26
2019-05-20 20.0 20.0 20.0 790.65 760.20 AXISBANK 782.40 -1.42 -3.95 74.52 752.09 745.98
2019-05-21 20.0 20.0 20.0 791.80 771.05 AXISBANK 772.95 0.64 -3.03 66.05 759.04 749.63
2019-05-22 20.0 20.0 20.0 784.00 770.20 AXISBANK 779.15 2.69 -1.88 68.74 765.75 753.62
2019-05-23 20.0 20.0 20.0 804.30 770.60 AXISBANK 776.40 4.07 -0.69 66.06 769.30 756.70
2019-05-24 20.0 20.0 20.0 795.90 778.50 AXISBANK 793.20 6.33 0.71 73.38 777.26 761.61
2019-05-27 20.0 20.0 20.0 816.40 787.20 AXISBANK 813.90 9.56 2.48 79.62 789.48 768.65
2019-05-28 20.0 20.0 20.0 822.10 801.00 AXISBANK 808.40 11.56 4.30 74.28 795.78 773.99
2019-05-29 20.0 20.0 20.0 813.00 802.10 AXISBANK 804.70 12.72 5.98 70.60 798.76 778.11
2019-05-30 20.0 20.0 20.0 814.20 803.25 AXISBANK 807.55 13.69 7.52 71.84 801.69 782.06
2019-05-31 20.0 20.0 20.0 818.70 793.20 AXISBANK 808.30 14.35 8.89 72.20 803.89 785.58
2019-06-03 20.0 20.0 20.0 814.65 803.10 AXISBANK 812.65 15.03 10.12 74.37 806.81 789.20
2019-06-04 20.0 20.0 20.0 827.75 805.50 AXISBANK 822.80 16.16 11.33 78.81 812.14 793.70
2019-06-06 20.0 20.0 20.0 825.95 805.80 AXISBANK 807.70 15.71 12.20 60.73 810.66 795.57
2019-06-07 20.0 20.0 20.0 813.25 801.30 AXISBANK 804.00 14.89 12.74 57.03 808.44 796.70
2019-06-10 20.0 20.0 20.0 816.75 807.40 AXISBANK 814.15 14.86 13.17 63.98 810.34 799.03
2019-06-11 20.0 20.0 20.0 821.45 807.00 AXISBANK 814.80 14.72 13.48 64.40 811.83 801.14
2019-06-12 20.0 20.0 20.0 815.75 809.10 AXISBANK 813.55 14.34 13.65 62.76 812.40 802.80
2019-06-13 20.0 20.0 20.0 823.40 810.75 AXISBANK 820.15 14.39 13.80 67.77 814.99 805.11
2019-06-14 20.0 20.0 20.0 823.10 797.45 AXISBANK 801.20 12.80 13.60 46.87 810.39 804.59
2019-06-17 20.0 20.0 20.0 801.00 775.10 AXISBANK 777.70 9.57 12.79 32.52 799.49 801.00
2019-06-18 20.0 20.0 20.0 784.85 770.10 AXISBANK 775.75 6.78 11.59 31.60 791.58 797.63
2019-06-19 20.0 20.0 20.0 789.00 763.35 AXISBANK 770.70 4.13 10.10 29.12 784.62 794.04
2019-06-20 20.0 20.0 20.0 775.45 758.85 AXISBANK 771.40 2.06 8.49 30.00 780.21 791.02
2019-06-21 20.0 20.0 20.0 778.25 762.80 AXISBANK 771.05 0.39 6.87 29.78 777.16 788.35
2019-06-24 20.0 20.0 20.0 775.25 761.00 AXISBANK 762.85 -1.57 5.18 24.99 772.39 784.95
2019-06-25 20.0 20.0 20.0 783.00 756.55 AXISBANK 781.65 -1.61 3.82 47.39 775.48 784.51
2019-06-26 20.0 20.0 20.0 792.00 781.70 AXISBANK 788.60 -1.08 2.84 53.33 779.85 785.06
2019-06-27 20.0 20.0 20.0 806.85 781.60 AXISBANK 800.45 0.29 2.33 61.82 786.72 787.11
2019-06-28 20.0 20.0 20.0 812.90 793.30 AXISBANK 808.55 1.99 2.26 66.61 793.99 789.97
2019-07-01 20.0 20.0 20.0 819.00 807.45 AXISBANK 810.45 3.45 2.50 67.71 799.48 792.70
2019-07-02 20.0 20.0 20.0 812.55 800.75 AXISBANK 802.60 3.93 2.79 58.54 800.52 794.02
2019-07-03 20.0 20.0 20.0 811.50 801.50 AXISBANK 806.20 4.55 3.14 61.31 802.41 795.64
2019-07-04 20.0 20.0 20.0 813.20 804.30 AXISBANK 808.85 5.20 3.55 63.39 804.56 797.41
2019-07-05 20.0 20.0 20.0 812.95 801.30 AXISBANK 806.10 5.42 3.93 59.56 805.07 798.57
2019-07-08 20.0 20.0 20.0 803.65 778.65 AXISBANK 782.90 3.70 3.88 37.50 797.68 796.48
2019-07-09 20.0 20.0 20.0 788.80 776.40 AXISBANK 786.20 2.57 3.62 41.08 793.85 795.11
2019-07-10 20.0 20.0 20.0 790.00 764.55 AXISBANK 769.25 0.32 2.96 30.67 785.65 791.66
2019-07-11 20.0 20.0 20.0 773.50 763.00 AXISBANK 766.40 -1.68 2.03 29.23 779.24 788.29
2019-07-12 20.0 20.0 20.0 768.60 751.00 AXISBANK 755.80 -4.07 0.81 24.33 771.42 783.96
2019-07-15 20.0 20.0 20.0 759.65 744.00 AXISBANK 749.75 -6.37 -0.62 21.92 764.20 779.40
2019-07-16 20.0 20.0 20.0 763.20 747.55 AXISBANK 761.05 -7.20 -1.94 35.69 763.15 776.95
2019-07-17 20.0 20.0 20.0 765.55 750.05 AXISBANK 752.60 -8.45 -3.24 30.97 759.63 773.70
2019-07-18 20.0 20.0 20.0 755.00 734.60 AXISBANK 740.40 -10.29 -4.65 25.39 753.22 769.26
2019-07-19 20.0 20.0 20.0 746.80 723.30 AXISBANK 729.25 -12.51 -6.22 21.33 745.23 763.93
2019-07-22 20.0 20.0 20.0 738.50 720.00 AXISBANK 727.45 -14.24 -7.83 20.71 739.30 759.06
2019-07-23 20.0 20.0 20.0 739.30 726.10 AXISBANK 728.00 -15.40 -9.34 21.51 735.54 754.92
2019-07-24 20.0 20.0 20.0 732.80 709.35 AXISBANK 712.30 -17.37 -10.95 16.15 727.79 749.24
2019-07-25 20.0 20.0 20.0 726.25 706.35 AXISBANK 723.15 -17.86 -12.33 30.06 726.24 745.76
2019-07-26 20.0 20.0 20.0 732.00 719.20 AXISBANK 729.85 -17.51 -13.37 37.45 727.45 743.64
2019-07-29 20.0 20.0 20.0 740.00 710.00 AXISBANK 720.25 -17.79 -14.25 31.88 725.05 740.52
2019-07-30 20.0 20.0 20.0 728.00 701.00 AXISBANK 707.60 -18.82 -15.17 26.00 719.23 736.13
2019-07-31 20.0 20.0 20.0 694.90 657.80 AXISBANK 674.10 -22.08 -16.55 16.63 704.19 727.86
2019-08-01 20.0 20.0 20.0 680.75 662.65 AXISBANK 669.30 -24.76 -18.19 15.69 692.56 720.05
2019-08-02 20.0 20.0 20.0 681.65 661.60 AXISBANK 673.85 -26.22 -19.80 20.58 686.32 713.89

Hi,
I am getting exception, when adx(df['close],]n=30,True) is called, please see attached file for dataframe.
adx_30_index_bound_error.txt

Traceback (most recent call last):
File "./algo_trade_signals.py", line 3499, in
do_identify_multibagger_on_monthly_charts(datetime_object_from_datetime,datetime_object_to_datetime,df)
File "./algo_trade_signals.py", line 2115, in do_identify_multibagger_on_monthly_charts
df['adx']=trend.adx(df['high'], df['low'], df['close'], n=30,fillna=True)
File "/usr/local/lib/python2.7/dist-packages/ta/trend.py", line 179, in adx
adx[n] = dx[0:n].mean()
IndexError: index 30 is out of bounds for axis 0 with size 11
Q2: why adx values initially for dataframe begining, zeros for more than 30 columns??

Hi,
i have a dataframe with daily stock market data for each EoD data i.e for each day one entry in dataframe.
when i calculate
rsi(n=30), rsi(n=14) on daily stock data, rsi(30) rsi(14) results same values almost similar values rsi(30) and rsi(14).
please see attached file for example results.
rsi_30_rsi_14_example_data.txt

Hi,

My dataset somehow posed a inf (infinitive) avlue in indicator volume_adi. Any reason why ? I tried stockstats and never had such weird error.

Thanks in advanced.
Steve

First: GREAT LIBRARY!!, @bukosabino!
(better than all the others!!)

And because of that I would be happy to find the parabolic sar from you in it. It's very complicated. I will try to code it for myself. If I'm ready with a good solution I will send it to you. But maybe you're faster! ;)

Many series return nan or inf on division by 0. It would be nice if a small epsilon was added so that the divisions are valid. Otherwise, setting fillna=True may mask some errors.

Thank you for all the work that you've put into this library. It is very useful! I was just using the ATR function and believe I found a minor error in the calculation. After calculating the TRs, the first ATR is set using the mean TR from the entire period which would not be able to be calculated point in time:

atr[0] = tr[1::].mean()

The effect of this only lasts for the length of the rolling window but shouldn't we set the first ATR value to just the first TR?

I emailed you an example as well.

Dear Developer,

Let me first express my gratitude for your project. I believe this is one of the best technical indicator library for python.

However, over the past couple of weeks, I have been having trouble with regression test. At first, the library had "ema_fast" and "ema_slow" functions, which were then replaced by a single "ema." This ema function had the default period of 14. However, the default value was removed as of version 2.

This results in code breaking down everytime the library is upgraded. This would make it difficult to use this library for a production code. I would like to suggest that you should not change the function signature outright, but rather show "deprecation warning" for a few versions first. So, people can migrate their codes.

Thank you very much.

for cci calculation, we should use "mean absolute deviation (mad)" not standard deviation. please correct it.
unfortunately, pandas rolling doesn't provide mad function. we have to find it by our own.

here I'm posting my code which is used for cci calculation.

pp = (data['CLOSE'] + data['LOW'] + data['HIGH'] )/3
mad = lambda x : np.mean(np.abs(x-np.mean(x)))
data['CCI']= (pp-pp.rolling(period).mean())/(0.015 * pp.rolling(period).apply(mad,True))

A few minor things:

1. The Ease function takes closing price as a positional argument, but the closing price isn't actually needed in the calculation.

2: The calculation itself is mostly correct, but is missing one step: It needs to be multiplied by 10^8 or 100,000,000.

3: The default n is 20, but I've only ever seen EMV calculated with n=14

df = pd.read_csv('D:/GLIM/Empirical/Data/NIFTY_ta.csv')

df.head()
Out[132]:
Date Open High Low Close Volume
0 10-Nov-08 2,973.30 3,161.25 2,973.30 3,148.25 25,07,00,000
1 11-Nov-08 3,147.20 3,147.20 2,919.45 2,938.65 22,16,54,000
2 12-Nov-08 2,937.90 2,975.20 2,794.95 2,848.45 25,04,60,000
3 14-Nov-08 2,848.00 2,938.80 2,778.80 2,810.35 25,67,42,000
4 17-Nov-08 2,813.40 2,835.70 2,694.50 2,799.55 25,12,20,000

Error 1

df2 = add_all_ta_features(df2, "Open", "High", "Low", "Close", "Volume", fillna=True)
Traceback (most recent call last):

File "", line 1, in
df2 = add_all_ta_features(df2, "Open", "High", "Low", "Close", "Volume", fillna=True)

File "D:\Setup Files\Anaconda3\lib\site-packages\ta\wrapper.py", line 185, in add_all_ta_features
df = add_volume_ta(df, high, low, close, volume, fillna=fillna)

File "D:\Setup Files\Anaconda3\lib\site-packages\ta\wrapper.py", line 26, in add_volume_ta
df[volume], fillna=fillna)

File "D:\Setup Files\Anaconda3\lib\site-packages\ta\volume.py", line 31, in acc_dist_index
clv = ((close - low) - (high - close)) / (high - low)

File "D:\Setup Files\Anaconda3\lib\site-packages\pandas\core\ops.py", line 1066, in wrapper
result = safe_na_op(lvalues, rvalues)

File "D:\Setup Files\Anaconda3\lib\site-packages\pandas\core\ops.py", line 1034, in safe_na_op
lambda x: op(x, rvalues))

File "pandas_libs\algos_common_helper.pxi", line 1212, in pandas._libs.algos.arrmap_object

File "D:\Setup Files\Anaconda3\lib\site-packages\pandas\core\ops.py", line 1034, in
lambda x: op(x, rvalues))

TypeError: unsupported operand type(s) for -: 'str' and 'str'

Error 2

df2 = add_all_ta_features(df2, df2['Open'], df2['High'], df2['Low'], df2['Close'], df2['Volume'], fillna=True)
Traceback (most recent call last):

File "", line 1, in
df2 = add_all_ta_features(df2, df2['Open'], df2['High'], df2['Low'], df2['Close'], df2['Volume'], fillna=True)

File "D:\Setup Files\Anaconda3\lib\site-packages\ta\wrapper.py", line 185, in add_all_ta_features
df = add_volume_ta(df, high, low, close, volume, fillna=fillna)

File "D:\Setup Files\Anaconda3\lib\site-packages\ta\wrapper.py", line 25, in add_volume_ta
df['volume_adi'] = acc_dist_index(df[high], df[low], df[close],

File "D:\Setup Files\Anaconda3\lib\site-packages\pandas\core\frame.py", line 2679, in getitem
return self._getitem_array(key)

File "D:\Setup Files\Anaconda3\lib\site-packages\pandas\core\frame.py", line 2723, in _getitem_array
indexer = self.loc._convert_to_indexer(key, axis=1)

File "D:\Setup Files\Anaconda3\lib\site-packages\pandas\core\indexing.py", line 1327, in _convert_to_indexer
.format(mask=objarr[mask]))

KeyError: "['3,161.25' '3,147.20' '2,975.20' ... '10,600.25' '10,616.45' '10,619.55'] not in index"

There is already a stochastic oscillator indicator, but can you also add Stochastic RSI indicator ? It would be of great help. I haven't found a single Stochastic RSI indicator, that matches with the one from Trading view.

EDIT: I updated my original answer because I think this might be better

Thanks for the hard work on the library. I noticed a bug in the calculation for On Balance Volume.

In the code:

` df = pd.DataFrame([close, volume]).transpose()

df['OBV'] = 0

c1 = close < close.shift(1)

c2 = close > close.shift(1)

if c1.any():

    df.loc[c1, 'OBV'] = - volume

if c2.any():

    df.loc[c2, 'OBV'] = volume

obv = df['OBV']

if fillna:

    obv = obv.replace([np.inf, -np.inf], np.nan).fillna(0)

return pd.Series(obv, name='obv')`

the lines:

if c1.any():

    df.loc[c1, 'OBV'] = - volume

if c2.any():

    df.loc[c2, 'OBV'] = volume`

don't seem to allow the OBV value to propagate through time. It is resetting obv = volume every time the close price is higher than the previous period, and obv = - volume every time the price closes lower.

If you initialize the OBV for period zero to be equal to volume at period zero then you should be able to apply a cumulative sum with cumsum . ex:

df.loc[0,'OBV'] = df.loc[0,'Volume']

Then assuming you keep the code you have, you can apply cumsum to what we currently have as the OBV column:

df.loc[:,'OBV']= df['OBV'].cumsum()

This should keep the cumulative sum as needed with as little effort as possible IMO.

Hi,
you can reach me on: [email protected]
When i read the documentation, and investopedia, i found that, even though indicators calculation is fine, but finding bullish divergences, bearish divergences, bullish failover, brearing failover are very and most important than, just finding techincal indicators.
please let me know if you need my help in python code development for the same.
I am a programmer from india, and very much liked your library.
i can contribute to your python lib, for divergences as a seperate library /method call for each technical indicator.

The dataframe I passed into the adx method had been resampled and the indexes were something like 60,120,180.

On line 144:
trs[0] = tr.dropna()[0:n].sum()

Due to my index starting at sixty, the series had no values after slicing and the sum was 0. This caused a divide by zero and created a nan. This caused din and dip to have the first value in the series be nan, which then resulted in all adx values being nan.

On line 145 there is a reset index. If this was moved prior to the sum for trs[0] it would resolve the issue. I'm not sure if there are other implications of doing that.

Hi!

First of all, really appreciate your effort in making TA calculation a lot easier in Python!

I just have one little request, can you add weighted moving average (WMA) in the library? Since simple moving average and exponential moving average are already in, WMA is also quite useful.

Hi,
I would like to calculate rsi for monthly performance, to get long term investment indicators using your rsi indicator.
Q: I would like to calculate rsi on daily stock market data with n=30 input parameter to your function: rsi(close, n=30, fillna=True),
will it work?.
Q: will this equiavalent to calculation of rsi(14) on monthly stock market data?
or what i mean is:
will rsi(30) on daily data in dataframe == rsi(14) on monthly data in dataframe?.

Encountered when running the pipeline example code.

pipe = make_pipeline(
TAFeaturesTransform("Open", "High", "Low", "Close", "Volume", fillna=True),
StandardScaler(),
RandomForestRegressor()
)

y_pred = pipe.fit(X_train, y_train).predict(X_test)

Hi,
Need your whatsapp , or facebook messenger to discuss about lstm and data science related.
Could you please share it?.
Regards,
Ravinder.Konka.
my what'sapp: +91 9603297252

In adx calculation you have:
pdm = high.combine(cs, lambda x1, x2: get_min_max(x1, x2, 'max'))
pdn = low.combine(cs, lambda x1, x2: get_min_max(x1, x2, 'min'))

The get_min_max function is not returning the max and min. I believe you forgot to add the returns, i.e.
return max(x1, x2)
return min(x1, x2)

@bukosabino

Hi

Please can you add CCI (Commodity Channel Index) indicator too?

If you don't know, there are three different Momentum base formulas in TA : MACD, TSI, and CCI.

You already have two of them why not adding the last one please?

Hi

I wanted to ask if the add_all_ta_features uses data from future candles or just current and past only
Thanks
P.S. Great work

Hi bukosabino,

As you are doing a great work, on ta using python.
Is it possible for you to make , a LSTM(Long short term memory model) using machine learning/data science to predict the stock market prices?.
I heard that, LSTM model is excellent model close to ~90% prediction rate for timeseries data of stocks.

Please let me know, if you need more information.

Hola,

First thank you for making this library available to the rest of us, second I have problem using the average_true_range function. As an example, I just tried to calculate the same values of the example you put as a reference:

https://stockcharts.com/school/doku.php?id=chart_school:technical_indicators:average_true_range_atr

But I couldn't get the same result as the refered page. I made sure my values were right, and finally I think the problem lies in the utils.ema function, and the way it calls pandas.ewm function.

I got it to produce the desired values, by calling the ewm function and specify the decay in terms of center of mass, like this:

return pd.concat([sma, rest]).ewm(com=(self.periods-1), adjust=False).mean()

instead of calling it like:

return pd.concat([sma, rest]).ewm(span=periods, adjust=False).mean()

why do you have the calculation of the average using the decay in terms of span?

the dpo is calculated as
close.shift(n/2+1) - close.sma(20)
however, according to the tutorial https://school.stockcharts.com/doku.php?id=technical_indicators:detrended_price_osci, the calculation should be
close - close.sma(20).shift(-(n/2+1))

Update:
Oh, I think they are the same thing, sorry!

Excellent job on developing all these indicators.
Can you please add KAMA - Kaufman's Adaptive Moving Average (https://school.stockcharts.com/doku.php?id=technical_indicators:kaufman_s_adaptive_moving_average) ?

It's an evolution of regular EMA (Esponential Moving Average).

I received an error when attempting to use the ADX feature using the code below:

df_data = pd.DataFrame()
df_data = pd.read_csv(text_response,names=fieldnames)
df_data= df_data.set_index('Date')
df_data['ADX']=ta.trend.adx('High','Low','Close',n=25,fillna=True)

These are the error messages that I received:

`
  File "<ipython-input-140-f9e022ae3a9c>", line 1, in <module>
    runfile('/home/tim/python/fx_related/fx_import.py', wdir='/home/tim/python/fx_related')

  File "/home/tim/anaconda3/lib/python3.6/site-packages/spyder_kernels/customize/spydercustomize.py", line 668, in runfile
    execfile(filename, namespace)

  File "/home/tim/anaconda3/lib/python3.6/site-packages/spyder_kernels/customize/spydercustomize.py", line 108, in execfile
    exec(compile(f.read(), filename, 'exec'), namespace)

  File "/home/tim/python/fx_related/fx_import.py", line 36, in <module>
    df_data['ADX']=ta.trend.adx('High','Low','Close',n=25,fillna=True)

  File "/home/tim/anaconda3/lib/python3.6/site-packages/ta/trend.py", line 137, in adx
    cs = close.shift(1)

AttributeError: 'str' object has no attribute 'shift'
```

Is anyone else having this problem?
Thanks

While installing I get following error -
I am on Python 2.7.10

pip install ta Collecting ta Using cached https://files.pythonhosted.org/packages/68/29/e2b9a4ad5f8710a0913d9fc1abb52c5b89ecf7ea036f8a8d8e4e73b08009/ta-0.1.4.tar.gz Complete output from command python setup.py egg_info: Traceback (most recent call last): File "<string>", line 20, in <module> File "/private/var/folders/53/3syytbw50fnc0stv_tz8s7qwmcg7vv/T/pip-build-ESKw_I/ta/setup.py", line 21, in <module> classifiers = [], File "/System/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/distutils/core.py", line 124, in setup dist.parse_config_files() File "/System/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/distutils/dist.py", line 390, in parse_config_files parser.read(filename) File "/System/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/ConfigParser.py", line 305, in read self._read(fp, filename) File "/System/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/ConfigParser.py", line 546, in _read raise e ConfigParser.ParsingError: File contains parsing errors: setup.cfg [line 2]: ' description-file = README.md\n'

Hey there!

While working with the library I dug through the library and found for instance for RSI that you just fill N/A values with 50:
momentum.py:48

if fillna:
        rsi = rsi.fillna(50)

is there any possible way to use pandas default fill-options like ffill on the df to just get the last known value if a new one cannot be established?

Hi I have problem to generate average_true_range.

It looks that there is problem in pandas.

I using python 3.6, pandas 0.23.4

(ta_2) c:\repos\ta_test\ta-master\examples_to_use>python all_features_example.py
Index(['Timestamp', 'Open', 'High', 'Low', 'Close', 'Volume_BTC',
'Volume_Currency', 'Weighted_Price'],
dtype='object')
Traceback (most recent call last):
File "all_features_example.py", line 17, in
fillna=True)
File "C:\Users\nxa18933\AppData\Local\Continuum\anaconda3\envs\ta_2\lib\site-packages\ta\wrapper.py", line 190, in add_all_ta_features
df = add_volatility_ta(df, high, low, close, fillna=fillna)
File "C:\Users\nxa18933\AppData\Local\Continuum\anaconda3\envs\ta_2\lib\site-packages\ta\wrapper.py", line 54, in add_volatility_ta
n=14, fillna=fillna)
File "C:\Users\nxa18933\AppData\Local\Continuum\anaconda3\envs\ta_2\lib\site-packages\ta\volatility.py", line 44, in average_true_range
atr[i] = (atr[i-1] * (n-1) + tr[i]) / n
File "C:\Users\nxa18933\AppData\Local\Continuum\anaconda3\envs\ta_2\lib\site-packages\pandas\core\series.py", line 767, in getitem
result = self.index.get_value(self, key)
File "C:\Users\nxa18933\AppData\Local\Continuum\anaconda3\envs\ta_2\lib\site-packages\pandas\core\indexes\base.py", line 3118, in get_value
tz=getattr(series.dtype, 'tz', None))
File "pandas_libs\index.pyx", line 106, in pandas._libs.index.IndexEngine.get_value
File "pandas_libs\index.pyx", line 114, in pandas._libs.index.IndexEngine.get_value
File "pandas_libs\index.pyx", line 162, in pandas._libs.index.IndexEngine.get_loc
File "pandas_libs\hashtable_class_helper.pxi", line 958, in pandas._libs.hashtable.Int64HashTable.get_item
File "pandas_libs\hashtable_class_helper.pxi", line 964, in pandas._libs.hashtable.Int64HashTable.get_item
KeyError: 3

Thank you for this package.

I found that kst implementation is different to
http://stockcharts.com/school/doku.php?id=chart_school:technical_indicators:know_sure_thing_kst

As for me, I'm adding another function which makes difference between kst and sig.
def kst_sig(close, r1=10, r2=15, r3=20, r4=30, n1=10, n2=10, n3=10, n4=15, nsig=9, fillna=False):
....
return pd.Series(kst-sig, name='kst-sig')

Hey this is the error I get, would be really happy if someone can help me, thankss

Collecting ta
Using cached https://files.pythonhosted.org/packages/ba/09/149cb10cbd6b98cbcbaa6f74169b4f1f2ce6df144335bea0df0eaa9e1b01/ta-0.4.2.tar.gz
Complete output from command python setup.py egg_info:
Traceback (most recent call last):
File "", line 1, in
File "/private/var/folders/hc/b2dnskds513b2dxjnv8m3g_40000gn/T/pip-install-ceehcvvi/ta/setup.py", line 14, in
install_requires=open('requirements.txt').readlines(),
FileNotFoundError: [Errno 2] No such file or directory: 'requirements.txt'

According to https://stockcharts.com/school/doku.php?id=chart_school:technical_indicators:ichimoku_cloud
for icichimoku cloud the values are shifted forward. Where can I get these values, are they discarded?

Hi!

Thanks for TA! Great job!
It seems that i'm getting inaccurate data with ADX, at least data is different compared to cryptowatch and other sources.
import krakenex
import ta
from pykrakenapi import KrakenAPI

api = krakenex.API()
k = KrakenAPI(api)
df, last = k.get_ohlc_data("XXBTZEUR", ascending=True, interval=5)
ohlc = ta.utils.dropna(df)
print(ta.trend.adx(ohlc.high, ohlc.low, ohlc.close))

and output is

...
2018-09-25 17:35:00    22.381099
Name: adx, Length: 720, dtype: float64

while to ADX on website is for same interval/lenght is 12.54

Hi, I got this error when I call function as below:

df = add_trend_ta(df, "high", "low", "close", fillna=False, colprefix="")

Thank you,
Ling

Hi
I have the whole 1 hour candle data of bitcoin from bitmex, you can't say they are wrong.
Then i replace nan with epsilon values
Collecting the data from index one forward
using the same code you showed in the read me for bollinger

the bollingers are,... anything but bollinger ! they are more like a spike every now and then between 0 and 1 !!!

http://prntscr.com/nbdmz4

the correct function call should be bollinger_lband for the bollinger bands themselves, maybe you could correct the ream me file. I understand that you are calculating something different for bollinger band indicator method. but showing an image of bollinger in read me, then putting those two function calls as example, will confuse users.

Also, a little file that showed all the available functions with their parameters as a reference would be nice. if you want, i can put near an hour and make that list for you

Just curious on what sort of requirements or expectations you have regarding the API now and moving forward.

• Limited to return a single Series? Is returning a DataFrame an option for indicators that return multiple series, such as MACD, BBANDs, et al?
• Are the indicators to be decoupled and not rely on others for computation? For instance, MACD relying on SMAs?

Thanks
KJ

I've seen an issue the trs[i] value can equal zero giving this error:

/usr/local/lib/python3.6/dist-packages/ta/trend.py:170: RuntimeWarning: invalid value encountered in double_scalars
  dip[i] = 100 * (dip_mio[i]/trs[i])
/usr/local/lib/python3.6/dist-packages/ta/trend.py:174: RuntimeWarning: invalid value encountered in double_scalars
  din[i] = 100 * (din_mio[i]/trs[i])

I've checked the pandas series sent to adx, and there are no np.nan, or None values in them.

It needs to be handled in some way. The code doesn't crash, it just gives a warning.

ta version 0.3.7

Hey there! I just fiddled around with your library and I really like it. One thing that feels kinda strange: when applying all indicators, the header just states "trend1" and so on, not the name of the actual indicator. This makes working with it quite tough.

Any chance that I missed something here?

Thank you!

I believe this not calculating the ADI correctly.

Per wikipedia the ADI should be ADI(prev) + volume X CLV
What those lines above are doing is (volume x CLV)(prev) + (volume x CLV)

So it's not propagating the previous ADI value to calculate the next ADI value, but rather the previous (volume x CLV) value.

Hey there, me again :D

So I ran some tests and watched the data and encountered some of the features giving me 0.0 as value. Some others suddenly lock in at strange fixed values also. (on the right in the screenshot, kcc at 7023.3 as an example)

I will attach a screenshot as well as the csv file that got created from my data.
(CSV is not supported by Github upload so I will attach a PDF)
output.pdf

I have seen this issue about seeing NaNs and 0.0s at the top of the returned dataframe when certain functions are used. Would it be possible to include the option to mitigate these NaNs and 0.0s? I am not referring to the fillna strategy, but rather something related to the calculation.
For example, I noticed that the cause of many of these NaNs and 0.0s are due to the use of pandas functions df.rolling(), df.shift() and df.diff(). The fixes would be to include parameters df.rolling(min_periods=0) and df.shift(fill_value=x) where x is some fill strategy like df.mean() or df.median(). Maybe this would not be a true representation of the calculation, but when using these functions for live data, they can provide some predictive power to any model as long as there are values (and not NaNs or 0.0s).
I can work on the implementation as this library has already proved extraordinarily useful and I would like to contribute. But I have not contributed to another project yet, so it might take some hand-holding.

Hey there,

Nice job with this library!

Can I ask why you have no (unit) tests to verify the correctness of the implementation of the indicators ?

I may be able to contribute there.

Cheers,

pip3 install ta succeeds.

>>> import pandas as pd
>>> from ta import *
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python3.6/dist-packages/ta/__init__.py", line 11, in <module>
    from .pipeline_wrapper import *
  File "/usr/local/lib/python3.6/dist-packages/ta/pipeline_wrapper.py", line 2, in <module>
    from sklearn.base import TransformerMixin, BaseEstimator
ModuleNotFoundError: No module named 'sklearn'

So it looks like the 'ta' package is missing sklearn as a dependency.

pip3 install sklearn

Python 3.6.7 (default, Oct 22 2018, 11:32:17) 
[GCC 8.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from ta import *
>>> 

Bollinger low and high calculate the standard deviation as follows:

mstd = close.rolling(n, min_periods=0).std()

According to the Pandas documentation, std() uses (N - ddof) as the denominator, where ddof defaults to 1, unlike numpy std().

StockCharts has an example of Bollinger in Excel which, unsurprisingly, uses N as the denominator. Wikipedia explains that N - 1 is used if the sd is being calculated from a sample of the population, rather than the whole. As this is rolling, and a very small set of numbers, I assume Pandas is calculating the actual sd from all the data fed to it, therefore we should use N.

The high and low band Bollinger calcs should be:

mstd = close.rolling(n, min_periods=0).std(ddof=0)

The same applies to the Bollinger indicators.

In your documentation, you have a link to MFI formula explanation and there you can see such text: "Money flow is positive when the typical price rises (buying pressure) and negative when the typical price declines (selling pressure)." So, you need to compare typical prices, not close prices.

df.loc[(df['Close'] > df['Close'].shift(1, fill_value=df['Close'].mean())), 'Up_or_Down'] = 1
df.loc[(df['Close'] < df['Close'].shift(1, fill_value=df['Close'].mean())), 'Up_or_Down'] = 2

Here you need to make a comparison with tp.

Hi sir,

this is a very nice open-library. highly appreciated.
One simple question, can we generate simple moving average with ta library in our python???

Thanks

The RSI calculated from Ta library is different from the one that you get from Trading View. I compared the results between Ta library and Talib library and , Talib gives you the same results as in tradingview.