qfinuwa 1.1.9.10

Framework for backtesting quantitative trading algorithims.

QFIN's Algorithmic Backtester (QFAB)

GitHub Page

Setup

To install on your system, use pip:

pip install qfinuwa

API Class

To pull market data ensure you have a text file with the API key and call API.fetch_stocks:

from qfinuwa import API

path_to_API = 'API_key.txt'
download_folder = './data'

API.fetch_stocks(['AAPL', 'GOOG', 'TSLA'], path_to_API, download_folder)

Indicator Class

Multi-Indicators

A multi-indicator takes in a single signal (price of an arbitary stock) and outputs a transformation of that stock.

It is called MultiIndicator because the indicator will have multiple values (one for each stock)

# Example 

class CustomIndicators(Indicators):
    
    @Indicators.MultiIndicator
    def bollinger_bands(self, stock: pd.DataFrame):
        BOLLINGER_WIDTH = 2
        WINDOW_SIZE = 100
        
        mid_price = (stock['high'] + stock['low']) / 2
        rolling_mid = mid_price.rolling(WINDOW_SIZE).mean()
        rolling_std = mid_price.rolling(WINDOW_SIZE).std()

        return {"upper_bollinger": rolling_mid + BOLLINGER_WIDTH*rolling_std,
                "lower_bollinger": rolling_mid - BOLLINGER_WIDTH*rolling_std}

Single-Indicators

Similar to MultiIndicator, SingleIndicator is implemented as a function that takes in stock data and returns an indicator or indicators.

It is called SingleIndicator because there is only a single signal.

# Example 
class CustomIndicators(Indicators):

    @Indicators.SingleIndicator
    def etf(self, stock: dict):

        apple = 0.2
        tsla = 0.5
        goog = 0.3

        return {'etf': apple*stock['AAPL'] + tsla*stock['TSLA'] + goog*stock['GOOG']}

Manually Testing

You can manually test you indicators as follows:

stock_a = pd.from_csv('stockA.csv')
stock_b = pd.from_csv('stockA.csv')

# multi-indicator for stockA (returns dict of dict of pd.Series)
output_a = CustomIndicators.bollinger(stockA)
# multi-indicator for stockB (returns dict of dict of pd.Series)
output_b = CustomIndicators.bollinger(stockA)

# single-indicator for stockA + stockB (returns dict of pd.Series)
output = CustomIndicators.etf({'stockA': stock_a, 'stockB': stock_b})

Hyper-parameters

Each function you implement can be thought of as a hyperparameter "group" that bundles the indicator it returns (the keys to the dictionary the indicator function returns).

The backtester can change hyperparameters for you, but to do so you need to give each one a name, in the form of kwargs.

The kwargs must include a default value which will be used if you do not specify a value.

class CustomIndicators(Indicators):
    
    @Indicators.MultiIndicator
    def bollinger_bands(self, stock: pd.DataFrame, BOLLINGER_WIDTH = 2, WINDOW_SIZE=100):
        
        mid_price = (stock['high'] + stock['low']) / 2
        rolling_mid = mid_price.rolling(WINDOW_SIZE).mean()
        rolling_std = mid_price.rolling(WINDOW_SIZE).std()

        return {"upper_bollinger": rolling_mid + BOLLINGER_WIDTH*rolling_std,
                "lower_bollinger": rolling_mid - BOLLINGER_WIDTH*rolling_std}

    @Indicators.SingleIndicator
    def etf(self, stock: dict, apple = 0.2, tsla= 0.5, goog=0.3):

        return {'etf': apple*stock['AAPL'] + tsla*stock['TSLA'] + goog*stock['GOOG']}

Strategy Class

To define your strategy extend qfin.Strategy to inherit its functionalities. Implement your own on_data function.

Your on_data function will be expected to take 4 positional arguments.

• self: reference to this object
• prices: a dictionary of numpy arrays of historical prices
• portfolio: object that manages positions

Similar to qfin.Indicators, you can define hyperparameters for your model in __init__.

# Example Strategy
class BasicBollinger(Strategy):

    def __init__(self, quantity=5):
        self.quantity = quantity
        self.n_failed_orders = 0
    
    def on_data(self, prices, indicators, portfolio):

        # If current price is below lower Bollinger Band, enter a long position
        for stock in portfolio.stocks:

            if(prices['close'][stock][-1] < indicators['lower_bollinger'][stock][-1]):
                order_success = portfolio.order(stock, quantity=self.quantity)
                if not order_success:
                    self.n_failed_orders += 1
            
            # If current price is above upper Bollinger Band, enter a short position
            if(prices['close'][stock][-1] > indicators['upper_bollinger'][stock][-1]):
                order_success = portfolio.order(stock, quantity=-self.quantity)
                if not order_success:
                    self.n_failed_orders += 1

    def on_finish(self):
        # Added to results object - access using result.on_finish
        return self.n_failed_orders

Additionally, you can specify a function on_finish that will run on the completion of a run, if you want to save your own data. Whatever this function returns will can be accessed in the results (see SingleRunResults.on_finish).

Backtester Class

The Backtester class asks for a custom strategy, custom indicators and data from the user. Once created, it can run multiple backtests without having to recalculate the indicators - when used in a Notebook environment the backtester object can persist and incrementally updated with new values.

Creating a Backtester

See qfinuwa.Backtester docstrings for specifics.

from qfinuwa import Backtester

backtester = Backtester(CustomStrategy, CustomIndicators, 
                        data=r'\data', days=90, 
                        delta_limits=1000, fee=0.01)

Updating Indicator Parameters

Update Parameters

backtester.indicators.update_params(dict_of_updates)

Get Current Parameters

backtester.indicators.params

Get Defaults

backtester.indicators.defaults

Updating Class

backtester.indicators = NewIndicatorClass

Updating Strategy Parameters

Update Parameters

backtester.strategy.update_params(dict_of_updates)

Get Current Parameters

backtester.strategy.params

Get Defaults

backtester.strategy.defaults

Updating Class

backtester.strategy = NewStrategyClass

Running a Backtester

Time Complexity Analysis

MIT License

Copyright (c) 2022 Isaac Bergl, QFIN UWA

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.