tools/ofi_delta.py

from utils.auxiliaries import create_dataframe
from scrapper.scrapper import Polygon
import pandas as pd
import numpy as np
import os


def calculate_ofi_and_delta(ticker, date, timeframe):

    def tick_rule(dataframe):
        return np.sign(dataframe["price"].diff().fillna(0).values)

    def order_flow_imbalance(group):
        group = quotes_dataframe.iloc[group.index]
        try:

            first_term = (group['bid_price'] >= group['bid_price'].shift(1)) * group['bid_size']
            second_term = (group['bid_price'] <= group['bid_price'].shift(1)) * group['bid_size'].shift(1)
            third_term = (group['ask_price'] <= group['ask_price'].shift(1)) * group['ask_size']
            fourth_term = (group['ask_price'] >= group['ask_price'].shift(1)) * group['ask_size'].shift(1)

            return pd.Series((first_term - second_term - third_term + fourth_term).sum())

        except KeyError:
            pass

    try:
        PATH = f'data\\{ticker}\\trade'
        file_name = f'{ticker}_trade_from_{date}_to_{date}.csv'
        trades_dataframe = pd.read_csv(os.path.join(PATH, file_name), index_col=0, parse_dates=True)
        trades_dataframe['participant_timestamp'] = pd.to_datetime(trades_dataframe['participant_timestamp'])

    except FileNotFoundError:
        trades_dataframe = Polygon().get_trades(ticker=ticker, from_date=date, to_date=date)

    try:
        PATH = f'data\\{ticker}\\quote'
        file_name = f'{ticker}_quote_from_{date}_to_{date}.csv'
        quotes_dataframe = pd.read_csv(os.path.join(PATH, file_name), index_col=0, parse_dates=True)
        quotes_dataframe['participant_timestamp'] = pd.to_datetime(quotes_dataframe['participant_timestamp'])

    except FileNotFoundError:
        quotes_dataframe = Polygon().get_quotes(ticker=ticker, from_date=date, to_date=date)

    trades_dataframe['type'] = tick_rule(trades_dataframe)
    trades_dataframe['type'] = trades_dataframe['type'].replace(0, np.nan).ffill()
    trades_dataframe = trades_dataframe.groupby([pd.Grouper(key='participant_timestamp', freq=timeframe)]).apply(create_dataframe)

    try:
        trades_dataframe = trades_dataframe.droplevel(1)
    except IndexError:
        pass

    trades_dataframe['OFI'] = quotes_dataframe.groupby([pd.Grouper(key='participant_timestamp', freq=timeframe)]).apply(order_flow_imbalance)
    trades_dataframe['return'] = trades_dataframe['close'].pct_change()
    trades_dataframe.index.name = 'date'

    return trades_dataframe