finlab-python / finlab_crypto Goto Github PK

class Strategy(object):

    def backtest(self, ohlcv, variables=dict(),
                         filters=dict(), lookback=None, plot=False,
                         signals=False, side='long', cscv_nbins=10,
                         cscv_objective=lambda r: r.mean(), html=None, compounded=True, execution_price='close', **args):

There might be a potential issue if using dictionary as the default argument for variables and filters, since dictionary is mutable.

I will argue that using None as the default argument might be a safer approach.

the colors red and green of the candlesticks render seems inverted. In uptrend are red, downtrend green I think should be the opposite.

For example like the pictures here: https://www.mt5traders.com/trend-candlesticks-metatrader-4-forex-indicator/

167 figures['entries & exits'] = pd.DataFrame(
168     {'entries':entries.squeeze(), 'exits': exits.squeeze()})
169 figures['performance'] = portfolio.cumulative_returns()

--> 171 c, info = chart.chart(ohlcv, overlaps=overlaps,
172 figures=figures, markerlines=mark_lines,
173 start_date=ohlcv.index[-min(1000, len(ohlcv))], end_date=ohlcv.index[-1], k_colors=k_colors)
174 c.load_javascript()
175 if html is not None:
...
161 # indicators
162 #################
164 def is_item(item):

TypeError: add_yaxis() got an unexpected keyword argument 'yaxis_data'

seems due to pyecharts library, you need to update your utility.py file to the new pyecharts version that receive 'y_axis' instead of 'yaxis_data'

ERROR: Cannot install finlab-crypto==0.1.0, finlab-crypto==0.1.1, finlab-crypto==0.1.10, finlab-crypto==0.1.11, finlab-crypto==0.1.12, finlab-crypto==0.1.13, finlab-crypto==0.1.14, finlab-crypto==0.1.15, finlab-crypto==0.1.16, finlab-crypto==0.1.17, finlab-crypto==0.1.18, finlab-crypto==0.1.19, finlab-crypto==0.1.2, finlab-crypto==0.1.3, finlab-crypto==0.1.4, finlab-crypto==0.1.5, finlab-crypto==0.1.6, finlab-crypto==0.1.7, finlab-crypto==0.1.8, finlab-crypto==0.1.9, finlab-crypto==0.2.0, finlab-crypto==0.2.1, finlab-crypto==0.2.10, finlab-crypto==0.2.11, finlab-crypto==0.2.12, finlab-crypto==0.2.13, finlab-crypto==0.2.14, finlab-crypto==0.2.15, finlab-crypto==0.2.16, finlab-crypto==0.2.17, finlab-crypto==0.2.18, finlab-crypto==0.2.19, finlab-crypto==0.2.2, finlab-crypto==0.2.20, finlab-crypto==0.2.21, finlab-crypto==0.2.22, finlab-crypto==0.2.23, finlab-crypto==0.2.3, finlab-crypto==0.2.4, finlab-crypto==0.2.5, finlab-crypto==0.2.6, finlab-crypto==0.2.7, finlab-crypto==0.2.8 and finlab-crypto==0.2.9 because these package versions have conflicting dependencies.
ERROR: ResolutionImpossible: for help visit https://pip.pypa.io/en/latest/topics/dependency-resolution/#dealing-with-dependency-conflicts

could you please provide an example how to use stop loss and trailing stop with backtest

Hello,

I am currently using Python 3.8.4 to run the tutorial with the finlab_crypto version 0.2.27.
I have encountered an issue where candlestick charts are not being displayed during the test.

Additionally, when running the optimization process, I encountered a TypeError: "pivot() takes 1 positional argument but 3 were given."

As a newcomer to Python, I would appreciate guidance on whether I need to install any additional dependencies or provide additional information to help identify the cause of these issues.

Thank you for your assistance.

__init__, __call__ , set_parameters and show_parameters methods looks similar in Strategy and Filter.
My suggestion is using Inheritance to refractor the code as following:

class BaseDec:

    delattr_clsname = {'Strategy'}

    def __init__(self, **default_parameters):
        self.func = None
        self._variables = None
        self.filters = {}
        self._default_parameters = default_parameters
        self.set_parameters(default_parameters)

    def __call__(self, func):
        self.func = func
        return self

    def set_parameters(self, variables):
        if type(self).__name__ in self.delattr_clsname:
          stop_vars = ['sl_stop', 'tp_stop', 'ts_stop']
          for svar in stop_vars:
              if hasattr(self, svar):
                  delattr(self, svar)

        if variables:
            for key, val in variables.items():
                setattr(self, key, val)

        self._variables = variables

    def show_parameters(self):
        print(self._variables)

class Filter(BaseDec):
    ...

class Strategy(BaseDec):
    ...

I would like to suggest including a .gitignore file in the repo.
We could use the template that github provided and modify it in the future if needed.

Stumbled upon this while researching vectorbt. I have few questions. hope you can answer

1. can this be used for stocks
2. How easy or difficult it is to integrate with alpaca for live trading

can u please give some guidance for me to code alpaca brokerage for stocks

Thanks for sharing this great repo!

I can't help but notice that the version of vectorbt is a bit outdated. I think it'd be very helpful to upgrade it to the latest 0.21.0, which offers enhanced stats and pnl analysis, and many more new features. I've created a pr #16 to support this. Hopefully you also find it useful. Thanks!

in function defined in crawler.py file row 100
get_nbars_binance(symbol, interval, nbars, client):
...
client (Binance.Client) (optional): Binance Client object.

It is not optional.
I've fixed with this: get_nbars_binance(symbol, interval, nbars, client=Client()):

In DH Bailey's paper "The probability of backtest overfitting", they used sharpe ratio as their default objective function.
However, in fin_crypto library, objective function was set up as objective=lambda r: r.mean().
Just want to clarify, needn't the default of objective function be set as objective=lambda r: sharpe_ratio(r)?

Great project!. A feature suggestion/request: cctx already standardizes data from many exchanges. Would be great to be able to use that data in order to have access to instruments supported on different exchanges (ex: futures, deribit options).