positions-backtester 0.1.2

Package to backtest trading strategies (dataframe with positions) with execution costs modeling

Table of Contents

• Short Overview.

• Installation via pip:

• How to use it

  • Create Backtester

  • How to backtest your dataframe with positions

  • Inputs:

    • df_positions_short

    • df_prices_full

  • Output: df_backtest_res

• Formulas

• Links

• Project local Links

• Contacts

• License

Short Overview.

positions_backtester is a python package (py>=3.7) to backtest trading strategies (dataframe with positions) with execution costs modeling

This package is trying to solve problem of slow trading with fast data.

Let’s say that you want to update your trading position once in a hour, day, week, …

But you have data with much higher resolution - minutes, seconds, miliseconds

Then you can give the wanted positions dataframe with the tick with which you want to trade

And higher resolution will be used to calculate approximate execution prices

Which will be just mean price over execution time period

Installation via pip:

pip install positions_backtester

How to use it

Create Backtester

from positions_backtester import Backtester

backtester = Backtester(float_percent_const_trading_fees=0.01,)

How to backtest your dataframe with positions

from positions_backtester import Backtester

df_backtest_res = backtester.backtest(
    df_positions_short,
    df_prices_full,
    is_to_neutralize=True,
    td_trading_delay=None,
    td_execution_duration=None,
)

Arguments:

1. df_positions_short:

   pd.DataFrame

   Positions we want to take with the frequency with which we want to change our positions

2. df_prices_full:

   pd.DataFrame

   Prices of assets in higher resolution

   Higher resolution needed for better execution evaluation

3. is_to_neutralize=True,:

   Flag if to have long-short equal positions

4. td_trading_delay=None:

   datetime.timedelta

   Delay needed to calculate the wanted positions

5. td_execution_duration:

   datetime.timedelta

   How long should the execution take

   Execution price will be the mean price over execution time period

Inputs:

df_positions_short

df_prices_full

Output: df_backtest_res

Formulas

PNL_before_costs = (previous_position) * (price_change_%)

trading_volume = abs(new_wanted_position - previous_position)

const_trading_fee_pnl = trading_volume * broker_commision

execution_fee_pnl = (new_wanted_position - previous_position) * (execution_price - current_price)

PNL_after_costs = PNL_before_costs - (const_trading_fee_pnl + execution_fee_pnl)

PNL_half_costs = PNL_before_costs - (const_trading_fee_pnl + execution_fee_pnl) / 2.0

Links

• PYPI

• GitHub

Project local Links

• CHANGELOG.

• CONTRIBUTING.

Contacts

• Email: [e-mail]

• vk.com

• Facebook

License

This project is licensed under the MIT License.