jrmeier/fast-trade

Error when run_backtest

uysalibov opened this issue · 9 comments

uysalibov commented 
print(validate_backtest(strategy_object))
datafile_path = "./archive/BTCUSDT1.csv"
result = run_backtest(strategy_object, datafile_path)

File "blablab\venv\lib\site-packages\fast_trade\build_data_frame.py", line 246, in standardize_df
new_df.open = pd.to_numeric(new_df.open)

AttributeError: 'DataFrame' object has no attribute 'open'

I cant run_backtest on my computer

jrmeier commented

Hi @uysalibov, can you verify the file /archive/BTCUSDT1.csv has the columns open,high, low,close as headers? It should look something like this sample csv file

uysalibov commented

Thanks for answer. Yes my csv file format doesn't have open, high, low and close headers i download your sample csv file and this time i get this error

File "bla\venv\lib\site-packages\fast_trade\build_data_frame.py", line 153, in apply_transformers_to_dataframe
    if len(ind.get("args", [])):
TypeError: object of type 'int' has no len()
jrmeier commented

gotcha. Can you verify all the indicators have an arguments property? If possible, can you post your strategy file? You can also run the validate_backtest function which might give more clues.

uysalibov commented

Yes, i validate_backtest. It returns

{'base_balance': None, 'chart_period': None, 'chart_start': None, 'chart_stop': None, 'comission': None, 'datapoints': None, 'enter': None, 'exit': None, 'any_enter': None, 'any_exit': None, 'trailing_stop_loss': None, 'exit_on_end': None, 'has_error': False}

Why it returns all keys None?

jrmeier commented

Hmm, there must be a bug where its missing this case. The None value should indicate any errors. In this case its not saying there's any errors. Can you confirm all your indicators have arguments and match the format in the README

uysalibov commented

I set all transformers from fast_trade.transformers_map
image

uysalibov commented

Can you check this is my example backtest?

{
    "name": "many-pandora",
    "base_balance": 100000,
    "exit_on_end": false,
    "chart_period": "15Min",
    "enter": [
        ["open", "<", "baspn__mid", 1],
        ["close", "<", "pivot__mid", 4],
        ["open", "=", "wma__short", 1],
        ["low", ">", "vzo__long", 2]
    ],
    "exit": [
        ["close", "=", "bbands__long", 1]
    ],
    "datapoints": [{
        "name": "baspn__mid",
        "transformer": "baspn",
        "args": 22
    }, {
        "name": "pivot__mid",
        "transformer": "pivot",
        "args": 44
    }, {
        "name": "wma__short",
        "transformer": "wma",
        "args": 21
    }, {
        "name": "vzo__long",
        "transformer": "vzo",
        "args": 23
    }, {
        "name": "bbands__long",
        "transformer": "bbands",
        "args": 41
    }]
}
jrmeier commented

Thanks for sharing. Inside of datapoints, each datapoint accepts args as a list, not a int. So changing for the datapoint baspn_mid it should look like

        "name": "baspn__mid",
        "transformer": "baspn",
        "args": [22] # should be a list of args

Hope this helps!

uysalibov commented

Yesss It works. Thanks for helping. 🎉

  • 👍1