eod_batch_runner.py

from risk_sensitivity import *
from simulation import *


def run_eod_batch():

    date_string = '2019-05-29'
    trade_date = pd.to_datetime(date_string)
    knowledge_time = pd.to_datetime('now')
    max_row = os.environ.get('max_row', '100')
    position_sims = os.environ.get('position_sims', '100')
    position_paths = os.environ.get('position_paths', '100')
    portfolio_paths = os.environ.get('portfolio_paths', '10')
    directory = os.environ.get('directory', ".")
    ins_df = get_instruments()
    books_df = get_books()
    cps_df = get_brokers()

    product_len = len(ins_df)
    books_len = len(books_df)
    cps_len = len(cps_df)

    write_reference = False
    if write_reference:
        print(ins_df)
        instruments = pa.Table.from_pandas(ins_df)
        pq.write_table(instruments, os.path.join(directory, 'instruments_eod.parquet'))

        print(books_df)
        books = pa.Table.from_pandas(books_df)
        pq.write_table(books, os.path.join(directory, 'books_eod.parquet'))

        print(cps_df)
        counterparties = pa.Table.from_pandas(cps_df)
        pq.write_table(counterparties, os.path.join(directory, 'counterparties_eod.parquet'))

    pos_df = None
    rows = []
    for i in range(int(max_row)):
        scale = 1
        if i % 988 == 0:
            scale = 25

        cur_row = {'PositionId': i,
                   'ProductId': randrange(product_len),
                   'BookId': randrange(books_len),
                   'CounterpartyId': randrange(cps_len),
                   'Quantity': random.randint(-100000, 100000)*scale,
                   'Exposure': random.randint(-100000, 100000)*scale,
                   'AsofDate': trade_date,
                   'KnowledgeTime': knowledge_time
                   }
        rows.append(cur_row)

        if i % 100000 == 0:
            pos_df = pos_df.append(pd.DataFrame(rows), ignore_index=True) if pos_df is not None else pd.DataFrame(rows)
            rows = []
            print('i is ' + str(i))

    pos_df = pos_df.append(pd.DataFrame(rows), ignore_index=True) if pos_df is not None else pd.DataFrame(rows)
    print(pos_df)

    if write_reference:
        positions = pa.Table.from_pandas(pos_df)
        pq.write_table(positions, os.path.join(directory, 'positions_eod.parquet'))


    calc_risk = False
    if calc_risk:
        write_credit_risk(max_row, directory)
        write_rate_risk(max_row, directory)
        write_vol_risk(max_row, directory)

    run_simulation = True
    if run_simulation:
        #write_portfolio_simulation(int(portfolio_paths), directory)
        write_position_simulation(int(position_sims), int(position_paths), directory)


if __name__ == "__main__":
    run_eod_batch()