TAlibIndicator

Adds one or more TAlib indicators to a data frame, based on existing open, high, low, and close column values.

TalibIndicator

Class Parameters

  • indicators
    • A list of indicators you want to transform the price information to.
  • lows
    • The lower end of the observation space. See spaces.Box to best understand.
  • highs
    • The lower end of the observation space. See spaces.Box to best understand.

Properties and Setters

  • NONE

Functions

Below are the functions that the TAlibIndicator uses to effectively operate.

Private

  • _str_to_indicator - Converts the name of an indicator to an actual instance of the indicator. For a list of indicators see list here.

Public

  • transform
    • Transform the data set and return a new data frame.

Use Cases:

Use Cases

Use Case #1: Selecting Indicators

The key advantage the TAlibIndicator has is that it allows us to dynamically set indicators according what’s inside of a list. For instance, if we’re trying to get the RSI and EMA together, we would run the following parameters inside ofthe indicator.

talib_indicator = TAlibIndicator(["rsi", "ema"])

This runs through the indicators in the list, at runtime and matches them to what is seen inside of TA-Lib. The features are then flattened into the output_space, both into the high and low segment of space.Box.

for i in range(len(self._indicators)):
    output_space.low = np.append(output_space.low, self._lows[i])
    output_space.high = np.append(output_space.high, self._highs[i])

Actual Use

from tensortrade.features import FeaturePipeline
from tensortrade.features.scalers import MinMaxNormalizer
from tensortrade.features.stationarity import FractionalDifference
from tensortrade.features.indicators import TAlibIndicator
price_columns = ["open", "high", "low", "close"]
normalize_price = MinMaxNormalizer(price_columns)
moving_averages = TAlibIndicator(["EMA", "RSI", "BB"])
difference_all = FractionalDifference(difference_order=0.6)
feature_pipeline = FeaturePipeline(steps=[normalize_price,
                                          moving_averages,
                                          difference_all])
exchange.feature_pipeline = feature_pipeline