PyPI version Build Status Coverage Status PyPI - Python Version PyPI - Wheel Documentation Status

DISCLAIMER: The results this backtesting software might produce may not be accurate, reliable or suppose any evidence ensuring the profitability of an algorithmic trading strategy. The results are indicative and might not be appropiate for trading purposes. Therefore, the creator does not bear any responsibility for any losses anyone might incur as a result of using this software.

Please, be fully informed about the risks and costs associated with trading the financial markets, as it is one of the riskiest investment forms possible.


alphabacktest is a library that aims at bringing algorithmic trading to all Pyhton programmers via a very simple set of methods that allow backtesting any trading strategy anyone can come up with, allowing external functions or modules.

You can find below an example of the structure to use when working with alphabacktest. The module is designed to be inherited in a new class created by the user and call the methods within the class. Backtest() has the engine to run the backtest chronologically and calls strategy() at each point in the data. The variables passed on to the function are list types with the previous quotes at each point for all the prices (high,close...) and volumes; except for dtime which is a string representing the current data point.

from alphabacktest import Backtest  

class mStrategy(Backtest):
    ''' Always call super().__init__() '''
    def __init__(self):

    ''' You can choose the parameters name'''
    def strategy(self,_open, close, high, low, vol, dtime):
        ''' Fill in your strategy '''
        if not self.has_positions():
            q = int(self.free_balance / close[-1])
            self.long_order(security=self.symbol, amount=q, dtime=dtime, price=close[-1])

if __name__ == '__main__':

As seen, the usage is pretty straightforward, and does not require a huge effort for the user to import and/or work with it, being a smooth process and giving freedom to apply any strategy, from those based on technical indicators to AI, going through many different tools.

Regarding the data source (explained at Usage), getting the proper data is sometimes rather difficult, especially when one is looking for tight timeframes (1m,5m,15m...). This data is not usually free, so this module gives the chance to either get the data from Yahoo Finance or from a csv file that the user can have in their local memory from a purchase or own harvest.


You can find all releases in PyPI.

$ pip install alphabacktest



As some of the features provided by the module are meant to be optional, the installation of alphabacktest does not imply the collection of TA-lib nor dash. Therefore, the package comes without these modules, which need to be installed by the user.

Usage ##

The super().__init__() is highly important as it defines exactly the settings of the backtest. It is customisable to an extent and allows multiple features to be desactivated or included.


Data source

In the first place, the data can be pulled from Yahoo Finance via pandas_datareader, or it can be imported from a local file.

From Yahoo Finance

In the first case, sym and file_path need to be left as they are, while the ticker needs to be filled with the security symbol (according to the symbols Yahoo Finance uses). The dateformat of Yahoo Finance is the default format.


From a local CSV file

If it is the second case, the initialization varies a little and sym, file_path and dateformat parameters need to be specified, and do not specify ticker. It is important that the dateformat is specified correctly, otherwise an error will be raised by the datetime module.

            dateformat="%Y-%m-%d %H:%M:%S",

However, the requirements for the file are very tight. The engine imports the data and assigns "Datetime","Open","High","Low","Close","Volume" values to the columns in that order. The format in Datetime column will be kept as a string and set as index whilst the format for Open-Volume will be converted to float. At last, the separator needs to be ','. If the format of your csv file is not this one, you can either transform it externally or provide your own data (which is recommended).

From a given DataFrame

The data format must be the same as the described above, meaning "Datetime" (str),"Open" (float),"High" (float),"Low" (float),"Close" (float),"Volume" (float) [IN THIS ORDER]. Datetime must be the dataframe's index.


data = pd.read_csv('csv/file/path.csv')
''' Data treatment'''
data= data.set_index("Datetime")
data.loc[:,'Open':'Close'] = data.loc[:,'Open':'Close'].astype(float)
class myStrategy(Backtest):
    def __init__(self,data):
        super().__init__(sym='BTCUSD',data=data,initial_time="04/01/2020 01:00:00",dateformat="%d/%m/%Y %H:%M:%S")

Data example:

                        Open     High      Low    Close  Volume
01/04/2007 17:15:00  3746.71  3850.91  3707.23  3843.52   4324200990.0
01/04/2007 17:30:00  3849.22  3947.98  3817.41  3943.41   5244856835.0
01/04/2007 17:45:00  3931.05  3935.69  3826.22  3836.74   4530215218.0
01/04/2007 18:00:00  3832.04  3865.93  3783.85  3857.72   4847965467.0
01/04/2007 18:15:00  3851.97  3904.90  4093.30  3836.90   5137609823.0
...                      ...      ...      ...      ...     ...

[m rows x n columns]


All the data available

If the aim of the user is to backtest the strategy for all the points in the data, the parameters initial_time and final_time should not be modified, they are already set as the very beginning and the last point respectively.

Specific timeframes

If the user wants to backtest certain scenarios, the initial_time and final_time need to be modified accordingly, always following the format specified in dateformat.

            initial_time="01/01/2015 00:00:00",
            final_time="01/01/2020 23:55:00",
            dateformat="%Y-%m-%d %H:%M:%S",


Technical Indicators

The technical indicators that come with the alphabacktest module are the SMA, EMA, RSI, Bollinger Bands and MACD. However if anyone would want to change it, you can also redefine the indicators() method.

from bcktclasses_cy import Backtest  

class mStrategy(Backtest):
    def __init__(self):

    def strategy(self,_open, close, high, low, vol, dtime):

        ''' Your strategy'''

    def indicators(self, close):
        from talib import RSI, BBANDS, MACD, EMA, SMA #....
        '''Define your indicators'''

if __name__ == '__main__':

Otherwise, if the user does not wish to use any of them, the indicators calculation can be desactivated by setting indicators=False in the super().__init() declaration.

Trading environment conditions

In order to simulate a real brokerage activity, the orders are not placed and executed right away. Instead, they first pass through a check and are placed in the following period. This adds more realism and settles the real-pessimistic scenario.

Moreover, the parameters that are required by the engine to reproduce this behaviour are set by default but the user can change them and adapt it to what their broker sets as conditions. The attributes are the following.

Results treatment

The default configuration is set to save the results in csv files inside a folder named backtest_results the engine creates in the cwd and it runs a Dash app where the results are plotted and the statistics presented. In case the user wanted to desactivate any of this features, the initialisation allows it by setting save_results or plot_results (bool) as False. Moreover, if the user wanted to change the directory where the results are stored, save_path (str) is the parameter to customize.



Class attributes

The attributes of the inherited class are the following.


        Security       OPrice                ODate CPrice CDate Amount       PNL Performance
btoHxGN    SP500  3653.884575  30/11/2020 20:30:00      -     -    -20  4.037615    0.110502


        Security       OPrice                ODate   CPrice                CDate Amount         PNL  Performance
l7h7eRx    SP500  3517.898175  12/10/2020 10:30:00  3482.75  14/10/2020 11:00:00    -20  699.445602    19.882486
eO9fL5Q    SP500  3444.844450  15/10/2020 04:00:00  3476.75  15/10/2020 14:30:00     20  634.666156    18.423652
sIWnkka    SP500  3503.649600  16/10/2020 09:45:00  3473.50  16/10/2020 15:00:00    -20  599.488350    17.110397
XihBLba    SP500  3423.092275  19/10/2020 14:00:00  3450.00  20/10/2020 05:30:00     20  534.731408    15.621297
Dm2U5jQ    SP500  3418.591825  26/10/2020 02:45:00  3396.25  26/10/2020 10:00:00     20 -450.255092   -13.170777
EAtb5UL    SP500  3388.088775  26/10/2020 10:30:00  3361.75  26/10/2020 12:30:00     20 -530.163589   -15.647866
3ua0S1x    SP500  3386.088575  26/10/2020 13:30:00  3360.00  27/10/2020 17:15:00     20 -525.157589   -15.509269


    Security Amount    Value
0    SP500    -20 -73070.0


       Security           Type             Datetime    Price Amount
lfdXIFo    SP500           Sell  12/10/2020 10:15:00  3516.00      1
6sp7cHv    SP500  Close#l7h7eRx  14/10/2020 10:45:00  3499.25    -20
JXKYklr    SP500            Buy  15/10/2020 03:45:00  3444.25      1
32Bc0fR    SP500  Close#eO9fL5Q  15/10/2020 14:15:00  3470.00     20
oP3w0Ha    SP500           Sell  16/10/2020 09:30:00  3507.00      1
XPTZCZE    SP500  Close#sIWnkka  16/10/2020 14:45:00  3484.00    -20
izB77xS    SP500            Buy  19/10/2020 13:45:00  3432.25      1
4qGx5Ld    SP500  Close#XihBLba  20/10/2020 05:15:00  3443.50     20


        Security           Type             Datetime    Price Amount     Status
DsN3JXP    SP500           Sell  12/10/2020 10:15:00  3516.00      1   Executed
Ztfajo7    SP500  Close#l7h7eRx  14/10/2020 10:45:00  3499.25    -20   Executed
VOiZwri    SP500            Buy  15/10/2020 03:45:00  3444.25      1   Executed
2hUedWm    SP500  Close#eO9fL5Q  15/10/2020 14:15:00  3470.00     20   Executed
O2j9Sjf    SP500           Sell  16/10/2020 09:30:00  3507.00      1   Executed
lStGqRL    SP500  Close#sIWnkka  16/10/2020 14:45:00  3484.00    -20   Executed
9lcGAFG    SP500            Buy  19/10/2020 13:45:00  3432.25      1   Executed
X86B0s7    SP500  Close#XihBLba  20/10/2020 05:15:00  3443.50     20   Executed

Class methods

The methods of the inherited class are the following ones.


The results are obtained via csv files and a dash app summary of the strategy performance that will be running on the local server as the logs report.

If the user wishes to plot the results after the backtesting on an independent script, the way to call the dash app is the following.

from alphabacktest._dash_results import DashApp


If you don't specify the dateformat, the default will be considered as %Y-%m-%d. For the system_path parameter, the default value is the cwd, otherwise it can be stated elsewhere, however all the results files need to be inside a folder called backtest_results.


TO DO: - Multiple assets (threading)


This package was created with Cookiecutter_ and the audreyr/cookiecutter-pypackage_ project template.

.. Cookiecutter: .. audreyr/cookiecutter-pypackage: