Digital currency of quantitative transactions - generate foot print factor data

import datetime
import time

import pandas as pd
from source_data.tqz_load_source_data import TQZSourceData

# from public_module.tqz_extern.tools.pandas_operator.pandas_operator import pandas

pd.set_option('mode.chained_assignment', None)


class TQZFootPrintTool:

    @classmethod
    def get_timestamp(cls, date_str: str):
        dt_str = f'{date_str} 08:00:00'
        s_timestamp = time.strptime(dt_str, "%Y-%m-%d %H:%M:%S")
        return int(time.mktime(s_timestamp) * 1000)


if __name__ == '__main__':
    symbol = 'BTCUSDT'
    start_date, end_date = datetime.datetime.strptime('2022-05-01', '%Y-%m-%d').date(), datetime.datetime.strptime('2022-05-02', '%Y-%m-%d').date()

    mergeTrades_data = TQZSourceData.load_mergeTrades_data(date_str=str(start_date), symbol=symbol)

    interval = 15 * 60000
    imbalance_price_ratio = 3
    qty_digital_point_counts = 3

    bar_start_timestamp = TQZFootPrintTool.get_timestamp(date_str=str(start_date))
    bar_end_timestamp = int(bar_start_timestamp) + interval

    daily_footPrint_df = pd.DataFrame(columns=["price", "timestamp", "bid_size", "ask_size", "bid_imbalance", "ask_imbalance"])
    while True:
        bar_trades_data = mergeTrades_data[(bar_start_timestamp <= mergeTrades_data['datetime']) & (mergeTrades_data['datetime'] < bar_end_timestamp)]

        # 1. create bar_price_size_df
        bar_price_size_df = pd.DataFrame(columns=["price", "bid_size", "ask_size", "bid_imbalance", "ask_imbalance"])
        for value in list(bar_trades_data.groupby('price')):
            bar_price_size_df = bar_price_size_df.append({
                'price': value[0],
                'ask_size': round(value[1][value[1]["isBuyerMaker"] == False]['qty'].sum(), qty_digital_point_counts),
                'bid_size': round(value[1][value[1]["isBuyerMaker"] == True]['qty'].sum(), qty_digital_point_counts)
            }, ignore_index=True)

        bar_price_size_df['ask_size_shift_-1'] = bar_price_size_df['ask_size'].shift(-1)
        bar_price_size_df['bid_size_shift_1'] = bar_price_size_df['bid_size'].shift(1)
        bar_price_size_df.dropna(subset=['ask_size_shift_-1', 'bid_size_shift_1'], inplace=True)

        bar_price_size_df.loc[bar_price_size_df['bid_size'] > (imbalance_price_ratio * bar_price_size_df['ask_size_shift_-1']), 'bid_imbalance'] = True
        bar_price_size_df.loc[(imbalance_price_ratio * bar_price_size_df['bid_size_shift_1']) < bar_price_size_df['ask_size'], 'ask_imbalance'] = True
        bar_price_size_df.fillna(False, inplace=True)

        bar_price_size_df.sort_values('price', ascending=False, inplace=True)
        bar_price_size_df.reset_index(inplace=True)
        del bar_price_size_df['index']
        del bar_price_size_df['ask_size_shift_-1']
        del bar_price_size_df['bid_size_shift_1']

        # 2. merge daily_footPrint_df & bar_price_size_df
        bar_price_size_df['timestamp'] = bar_start_timestamp
        daily_footPrint_df = pd.concat([daily_footPrint_df, bar_price_size_df], axis=0)

        # 3. update bar_start_timestamp & bar_end_timestamp at last.
        bar_start_timestamp = bar_end_timestamp
        bar_end_timestamp = int(bar_start_timestamp) + interval

        # 4. next day.
        if str(pd.to_datetime(f'{bar_start_timestamp}', unit='ms').date()) != str(start_date):
            # 4.1 write foot_print.csv.
            daily_footPrint_df.to_csv(f'F:/tqz_test/{symbol}_{str(int(interval / 60000))}mBar_footPrint_{str(start_date)}.csv', index=False)

            # 4.2 change start_date & mergeTrades_data & daily_footPrint_df.
            start_date += datetime.timedelta(days=1)
            mergeTrades_data = TQZSourceData.load_mergeTrades_data(date_str=str(start_date), symbol=symbol)
            daily_footPrint_df = pd.DataFrame(columns=["price", "timestamp", "bid_size", "ask_size", "bid_imbalance", "ask_imbalance"])

            if start_date > end_date:
                break