Skopt optimizer issues with VS Code and Jupyter notebooks and or possibly file structure?

[eervin123]

Hi all, I run VS Code on various machines, Work- windows 10 pc, home-Mac.

I have been using and really appreciating the newer skopt functionality with the optimize function. However, recently I came across an issue which I believe is operator error, but I can't get to the bottom of it. I didn't want to post it as an issue because I don't think it is a problem with the library.

Folder Structure:
.
+-- optimizer_notebook.ipynb
+-- scripts
| +-- init.py
| +-- ember_strategies.py

The main.ipynb has the following

from backtesting import Backtest, Strategy
from backtesting.test import GOOG
from backtesting.lib import resample_apply, OHLCV_AGG
from numpy import arange
from scripts.ember_strategies import (
    Benchmark,
    simpleSmaCross,
)

# run a backtested strategy
bt = Backtest(
    GOOG,
    simpleSmaCross,
    margin=1,
    cash=100000,
    commission=0.002,
    exclusive_orders=False,
)
stats = bt.run()
bt.plot()

Everything works fine here.

Then I run the following two cells

%%capture
! pip install scikit-optimize  # This is a run-time dependency

and

%%time
stats_skopt, heatmap, optimize_result = bt.optimize(
    n1=[10, 100],      # Note: For method="skopt", we
    n2=[20, 200],      # only need interval end-points
    constraint=lambda p: p.n1 < p.n2,
    maximize='Equity Final [$]',
    method='skopt',
    max_tries=200,
    random_state=0,
    return_heatmap=True,
    return_optimization=True)

and I get the following error.

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<timed exec> in <module>

~\Anaconda3\lib\site-packages\backtesting\backtesting.py in optimize(self, maximize, constraint, return_heatmap, **kwargs)
   1281                                   "set multiprocessing start method to 'fork'.")
   1282                 for batch_index in _tqdm(range(len(param_batches))):
-> 1283                     _, values = Backtest._mp_task(backtest_uuid, batch_index)
   1284                     for value, params in zip(values, param_batches[batch_index]):
   1285                         heatmap[tuple(params.values())] = value

~\Anaconda3\lib\site-packages\backtesting\backtesting.py in _mp_task(backtest_uuid, batch_index)
   1306         return batch_index, [maximize_func(stats) if stats['# Trades'] else np.nan
   1307                              for stats in (bt.run(**params)
-> 1308                                            for params in param_batches[batch_index])]
   1309 
   1310     _mp_backtests = {}  # type: Dict[float, Tuple[Backtest, List, Callable]]

~\Anaconda3\lib\site-packages\backtesting\backtesting.py in <listcomp>(.0)
   1304     def _mp_task(backtest_uuid, batch_index):
   1305         bt, param_batches, maximize_func = Backtest._mp_backtests[backtest_uuid]
-> 1306         return batch_index, [maximize_func(stats) if stats['# Trades'] else np.nan
   1307                              for stats in (bt.run(**params)
   1308                                            for params in param_batches[batch_index])]

~\Anaconda3\lib\site-packages\backtesting\backtesting.py in <genexpr>(.0)
   1306         return batch_index, [maximize_func(stats) if stats['# Trades'] else np.nan
   1307                              for stats in (bt.run(**params)
-> 1308                                            for params in param_batches[batch_index])]
   1309 
   1310     _mp_backtests = {}  # type: Dict[float, Tuple[Backtest, List, Callable]]

~\Anaconda3\lib\site-packages\backtesting\backtesting.py in run(self, **kwargs)
   1110         data = _Data(self._data.copy(deep=False))
   1111         broker = self._broker(data=data)  # type: _Broker
-> 1112         strategy = self._strategy(broker, data, kwargs)  # type: Strategy
   1113 
   1114         strategy.init()

~\Anaconda3\lib\site-packages\backtesting\backtesting.py in __init__(self, broker, data, params)
     57         self._broker = broker  # type: _Broker
     58         self._data = data   # type: _Data
---> 59         self._params = self._check_params(params)
     60 
     61     def __repr__(self):

~\Anaconda3\lib\site-packages\backtesting\backtesting.py in _check_params(self, params)
     75                     "Strategy '{}' is missing parameter '{}'. Strategy class "
     76                     "should define parameters as class variables before they "
---> 77                     "can be optimized or run with.".format(self.__class__.__name__, k))
     78             setattr(self, k, v)
     79         return params

AttributeError: Strategy 'simpleSmaCross' is missing parameter 'method'. Strategy class should define parameters as class variables before they can be optimized or run with.

However, in looking at my ember_strategies.py

from types import DynamicClassAttribute
from numpy.lib.arraypad import pad
import pandas as pd
from backtesting.lib import (
    SignalStrategy,
    TrailingStrategy,
    resample_apply,
    crossover,
    barssince,
)
from backtesting.test import SMA
from backtesting import Backtest, Strategy


class Benchmark(Strategy):  # Buy and Hold for calculating Statistics
    def init(self):
        self.buy()

    def next(self):
        pass


class simpleSmaCross(Strategy):
    n1 = 50
    n2 = 200

    def init(self):
        self.sma1 = self.I(SMA, self.data.Close, self.n1)
        self.sma2 = self.I(SMA, self.data.Close, self.n2)

    def next(self):
        if crossover(self.sma1, self.sma2):
            self.buy()
        elif crossover(self.sma2, self.sma1):
            # self.sell()
            self.position.close()

I blame this on me or my VS Code settings because when I run virtually the exact same code and folder structure in a .py file everything works fine, when I run it on my Mac at home, it works fine in both .ipynb as well as .py and when I create a brand new project in VS Code and build something that I think is perfectly identical on the PC, it works fine. Also, I know that it is importing the strategy class paramaters because when I run:
simpleSmaCross.__dict__ in the notebook I get the following:

mappingproxy({'__module__': 'scripts.ember_strategies',
              'n1': 50,
              'n2': 200,
              'init': <function scripts.ember_strategies.simpleSmaCross.init(self)>,
              'next': <function scripts.ember_strategies.simpleSmaCross.next(self)>,
              '__doc__': None,
              '__abstractmethods__': frozenset(),
              '_abc_impl': <_abc_data at 0x1e16d3be7b0>})

which I am assuming implies that it has the necessary parameters.

I have seen things on the web that suggest VS Code and Jupyter extension can be tricky when it comes to folder structure, and packages and module imports, but the error message is what made me want to post the question here.

Strategy 'simpleSmaCross' is missing parameter 'method'. Strategy class should define parameters as class variables before they can be optimized or run with.

what is the parameter 'method' referring to?

BTW, everything was working fine for optimizations in this project on this PC a week or so ago. My thinking is that somehow my Mac at home has a VS code settings file that accidentally was included in my github repo when I pulled changes to the windows machine.

Any help would be appreciated.

Python Version 3.8
backtesting version 3.0

[kernc]

Strategy 'simpleSmaCross' is missing parameter 'method'. Strategy class should define parameters as class variables

method='skopt' is interpreted as **kwargs strategy parameters. You're likely not really using Backtesting 0.3.0. Traceback snippets confirm it.

pip install -U backtesting

[eervin123]

I didn't get the method='skopt' being the method, so thanks. I could swear I checked that I was using the proper version of backtesting but it could be some environment issue that I thought I was in one environment with the proper backtesting and was really in another one. Thanks. I'll test it out on Monday and let you know what I learn.

[eervin123]

Problem solved, it had something to do with my configuration about jupyter extension in vscode pulling a different environment than what was selected in the terminal. Thanks for clearing up the confusion.