TALIpp - Incremental Technical Analysis Library

talipp (or tali++) is a Python library implementing financial indicators for technical analysis. The distinctive feature of the library is its incremental computation which fits extremely well real-time applications or applications with iterative input in general.

Unlike existing libraries for technical analysis which typically have to work on the whole input vector in order to calculate new values of indicators, talipp due to its incremental architecture calculates new indicators' values exclusively based on the delta input data. That implies, among others, it requires O(1) time to produce new values in comparison to O(n) (or worse) required by other libraries.

Supported incremental operations include:

appending new values to the input
updating the last input value
removing arbitrary number of the input values

Besides the already mentioned superior time complexity for delta input operations, talipp's incremental approach immediately offers other interesting features for free, such as indicator chaining or building new indicators combined from other indicators. See section with examples to get an idea.

Incremental nature of talipp naturally excels in applications with frequent CUD operations but it can be used for charting, back-testing, ... as any other existing library.

Last but not least, talipp is a very young project and therefore open to any suggestions of amending the API to users' liking. You are encouraged to come up with proposals.

What's new in version 1.7.0

OHLCV class supports timestamps

For the full history of changes see CHANGELOG.

List of incremental indicators

talipp currently provides below set of indicators. If your favourite indicator is missing, then create a ticket via GitHub Issues and there is a good chance that it will be included in the future version of the library.

Accumulation/Distribution (ADL)
Aroon
Average Directional Index (ADX)
Average True Range (ATR)
Awesome Oscillator (AO)
Balance of Power (BOP)
Bollinger Bands (BB)
Chaikin Oscillator
Chande Kroll Stop
Choppiness Index (CHOP)
Coppock Curve
Commodity Channel Index (CCI)
Donchian Channel (DC)
Detrended Price Oscillator (DOP)
Ease of Movement (EMV)
Force Index
Ichimoku Kinko Hyo
Keltner Channel (KC)
Klinger Volume Oscillator (KVO)
Know Sure Thing (KST)
Mass Index
McGinley Dynamic
Mean Deviation
Moving Average (ALMA, SMA, SMMA, DEMA, EMA, HMA, KAMA, TEMA, VWMA, WMA)
Moving Average Convergence Divergence (MACD)
On-balance Volume (OBV), Smoothed On-balance Volume (SOBV)
Parabolic SAR
Pivots High/Low
Rate of Change (ROC)
Relative strength index (RSI)
SFX TOR
Standard Deviation
Stochastic Oscillator
Stochastic RSI
TRIX
True Strength Index (TSI)
Ultimate Oscillator (UO)
Vortex Indicator (VTX)
Volume Weighted Average Price (VWAP)

Installation

pip install talipp

In case you want to install the latest version from the repo, use

pip install git+https://github.com/nardew/talipp.git@master

Examples

Consult examples folder to see usage of every single indicator included in the library. To get the basic look and feel of the API, see below.

from talipp.indicators import EMA, SMA, Stoch
from talipp.ohlcv import OHLCVFactory

# EMA indicator ([float] -> [float])
ema = EMA(period = 3, input_values = [1, 3, 5, 7, 9, 2, 4, 6, 8, 10])

# treat indicators as any other list
print(f'EMA(3): {ema}') # [3.0, 5.0, 7.0, 4.5, 4.25, 5.125, 6.5625, 8.28125]
print(f'Last EMA value: {ema[-1]}') # 8.28125

# append a new input value incrementally
ema.add_input_value(11)
print(f'EMA after adding a new value:      {ema}') # [3.0, 5.0, 7.0, 4.5, 4.25, 5.125, 6.5625, 8.28125, 9.640625]

# change the last added value
ema.update_input_value(15)
print(f'EMA after updating the last value: {ema}') # [3.0, 5.0, 7.0, 4.5, 4.25, 5.125, 6.5625, 8.28125, 11.640625]

# change the last added value again
ema.update_input_value(18)
print(f'EMA after updating the last value: {ema}') # [3.0, 5.0, 7.0, 4.5, 4.25, 5.125, 6.5625, 8.28125, 13.140625]

# remove the last added value
ema.remove_input_value()
print(f'EMA after removing the last value: {ema}') # [3.0, 5.0, 7.0, 4.5, 4.25, 5.125, 6.5625, 8.28125]

# purge the oldest input value
ema.purge_oldest(1)
print(f'EMA after purging the oldest value: {ema}') # [5.0, 7.0, 4.5, 4.25, 5.125, 6.5625, 8.28125]

# STOCH indicator ([OHLCV] -> [composite])
stoch = Stoch(5, 3, OHLCVFactory.from_dict({
    'high':     [5, 10, 15, 20, 25, 30, 35],
    'low':      [1, 4, 7, 10, 13, 16, 19],
    'close':    [3, 9, 8, 19, 18, 17, 19]
}))

# print result as a list of composite values for 'k' and 'd' output parameters
print(f'Stoch(5, 3) composite result: {stoch}') # [StochVal(k=70.83333333333333, d=None), StochVal(k=50.0, d=None), StochVal(k=42.857142857142854, d=54.563492063492056)]

# print result as lists per output parameters
print(f'Stoch(5, 3) decomposed result: {stoch.to_lists()}') # {'k': [70.83333333333333, 50.0, 42.857142857142854], 'd': [None, None, 54.563492063492056]} 

# Indicator chaining
sma1 = SMA(3)
sma2 = SMA(3, input_indicator = sma1)
sma3 = SMA(3, input_indicator = sma2)

print(f"Chain three moving averages:")
sma1.add_input_value([1, 2, 3, 4, 5, 6, 7, 8, 9, 10])
print(f"SMA1: {sma1}") # [2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0]
print(f"SMA2: {sma2}") # [3.0, 4.0, 5.0, 6.0, 7.0, 8.0]
print(f"SMA3: {sma3}") # [4.0, 5.0, 6.0, 7.0]

print(f"Purge oldest 3 values:")
sma1.purge_oldest(3)
print(f"SMA1: {sma1}") # [5.0, 6.0, 7.0, 8.0, 9.0]
print(f"SMA2: {sma2}") # [6.0, 7.0, 8.0]
print(f"SMA3: {sma3}") # [7.0]

Performance

To illustrate performance scaling of talipp we ran several tests together with the industry standard talib library and its python wrapper ta-lib. The takeaway from the comparison is following:

for batch processing (i.e. one-off calculation of indicators without addition of further delta values) talib is a clear winner. This is not surprising at all since it is implemented in C and it is tailored for vector calculations in one shot. talipp's incremental (i.e. not vector) calculation and features such as indicator chaining (which internally implements output listeners) must inevitably come at a cost. That being said, talipp calculates SMA for batch of 50k values incrementally still in ~200ms which is perfectly acceptable for many applications
where talipp clearly takes the lead is incremental calculation. Again this is well expected since talipp's CUD operations take O(1) time compared to O(n) time of talib. For 50k input the difference is as big as ~200ms vs. ~6800ms.
from the graphs it is apparent that talipp scales linearly with the size of the input compared to quadratic curve of talib when incremental operations are concerned. This follows from talipp's O(1) time for delta operations vs. talib's O(n).

Contact

to report issues, bugs, corrections or to propose new features use preferably Github Issues
for topics requiring more personal approach feel free to send an e-mail to

Support

If you like the library and you feel like you want to support its further development, enhancements and bug fixing, then it will be of great help and most appreciated if you:

file bugs, proposals, pull requests, ...
spread the word
donate an arbitrary tip
- BTC: 3GJPT6H6WeuTWR2KwDSEN5qyJq95LEErzf
- ETH: 0xC7d8673Ee1B01f6F10e40aA416a1b0A746eaBe68
- Binance Smart Chain tokens: 0xe37FaB52ed4c1C9a3d80896f2001Cb3284a1b619
- XMR: 87vdCaWFN2YJEk3HKVJNaPBFsuwZTJocRfpGJ747dPQrFcrs6SQTmA3XDGyWUPoALuNnXezEbJXkbY8Y4VSxG4ReEFqxy5m

SaeidRp/talipp