@guyue We can't help you without a clear question

Still digging... looks like this deal is actually the one which may be causing the issue...

It's somehow placing the sell before the first buy:

*** NEW DEAL *** BUYING: size 0.007998528270798173 price 1250.23 DEBUGGING: Size 0 Take Profit Price 0.0 2021-01-07T13:26:00, SELL EXECUTED, Size: -0.04059 Price: 1261.41, Cost: -51.20, Comm 0.00 2021-01-07T13:26:00, BUY EXECUTED, Size: 0.00800 Price: 1250.61, Cost: -10.09, Comm 0.00 DEBUGGING: Size 0.007998528270798173 Take Profit Price 1266.2426249999999 2021-01-07T16:39:00, SELL EXECUTED, Size: -0.00800 Price: 1265.86, Cost: -10.12, Comm 0.00

Given how I have my def next() setup, I am unsure how this is possible...

@run-out I thought I had this working with backtrader.analyzers.PyFolio() however I then realized it is not meant to handle 1m candles (or anything < 1d), so although the report was populating, the data seemed incorrect.

Is it possible to load data into quanstats reports using 1m candles? I am trying to do it with backtrader.analyzers.TimeReturn() but running into some issues... and before I kept digging wanted to see if this is the right approach.

def backtest_pair(pair, tp, bo, so, mstc, os, sos, ss): # Create a cerebro entity cerebro = bt.Cerebro() # Add the Data Feed to Cerebro data = load_data(pair) cerebro.adddata(data) # Add Strategy cerebro.addstrategy(DCA3C, tp=tp, bo=bo, so=so, mstc=mstc, os=os, sos=sos, ss=ss) # Add Analyzer cerebro.addanalyzer(bt.analyzers.TimeReturn, _name='Returns', timeframe=bt.TimeFrame.Minutes, compression=1) # Set starting cash starting_cash = calc_max_funds(bo, so, mstc, os) cerebro.broker.setcash(starting_cash) # Set the commission - 0.1% ... divide by 100 to remove the % cerebro.broker.setcommission(commission=0.001) print('Starting Portfolio Value: %.2f' % cerebro.broker.getvalue()) # Run results = cerebro.run() # cerebro.plot() strat = results[0] portfolio_stats = strat.analyzers.getbyname('Returns') returns = pd.DataFrame.from_dict(portfolio_stats.get_analysis(), orient="index") returns.index = pd.to_datetime(returns.index) quantstats.extend_pandas() # returns.index = returns.index.tz_convert(None) # Save bot_settings = 'TP: <b>{}</b> BO: <b>{}</b> SO: <b>{}</b> MSTC: <b>{}</b> OS: <b>{}</b> SOS: <b>{}</b> SS: <b>{}</b>'.format( strat.params.tp, strat.params.bo, strat.params.so, strat.params.mstc, strat.params.os, strat.params.sos, strat.params.ss) quantstats.reports.html(returns, output=f'{pair.replace("/", "_")}.html', title=f'Pair: <b>{pair}</b> {bot_settings}<br>', trading_year_days=365) print('Final Portfolio Value: %.2f' % cerebro.broker.getvalue())

Getting this error:

Traceback (most recent call last): File "/Users/ethan/PycharmProjects/backtraderbot/main.py", line 315, in <module> backtest_pair('ETH/USDT', tp=1.75, bo=10, so=20, mstc=30, os=1.05, sos=2, ss=1) File "/Users/ethan/PycharmProjects/backtraderbot/main.py", line 250, in backtest_pair trading_year_days=365) File "/Users/ethan/PycharmProjects/backtraderbot/venv/lib/python3.7/site-packages/quantstats/reports.py", line 101, in html by='max drawdown', ascending=True)[:10] File "/Users/ethan/PycharmProjects/backtraderbot/venv/lib/python3.7/site-packages/pandas/util/_decorators.py", line 311, in wrapper return func(*args, **kwargs) File "/Users/ethan/PycharmProjects/backtraderbot/venv/lib/python3.7/site-packages/pandas/core/frame.py", line 6254, in sort_values k = self._get_label_or_level_values(by, axis=axis) File "/Users/ethan/PycharmProjects/backtraderbot/venv/lib/python3.7/site-packages/pandas/core/generic.py", line 1776, in _get_label_or_level_values raise KeyError(key) KeyError: 'max drawdown'

Thanks!

@run-out Thanks!

@vladisld much appreciated mate :) That worked out

UPDATE:
Solved it by defining trading hours in calendar

class MyCalendar(bt.TradingCalendar): params = { 'open': time(9, 15), 'close': time(15, 30), }

Thanks for your help! This is a popular and well-tested trading strategy. And it's one of the best crypto trading methods for beginners. bitcoin news - here you can read more about btc trading strategies.

@hghhgghdf-dfdf

Can you please how your backtesting P/L looks like. I'm assuming we buy at 21.84 and sell at 15.90 is it? are there any other confirmation studies we can add to it ? as mostly Zig Zag repaints(overwrites the signals)

@run-out Thanks! I think your idea works, but that is what I want.

before I start to try backtrader, I use talib+pandas, they are easy to transfer signals from :

signalA=[0,0,1,0,0,0,0,-1,0,0] # triggered at 3rd day,not yet signalB=[0,0,0,1,0,0,-1,0,0,0] # triggered at 4rd day,not yet signalC=[0,0,0,0,0,1,0,-1,0,0] # triggered at 6rd day, buy here

to

signalA=[0,0,1,1,1,1,1,-1,-1,-1] # triggered at 3rd day,not yet signalB=[0,0,0,1,1,1,-1,-1,-1,-1] # triggered at 4rd day,not yet signalC=[0,0,0,0,0,1,1,-1,-1,-1] # triggered at 6rd day, buy here

then , when signalA==1 and signalB==1 and signalC==1, it trigger the long signal.
Since I am a newbie for backtrader, I don't know to datafeeds work that way.

@run-out It works, thanks!!!

@toinou222 instead of modifying the commission scheme you can simply increment the value of the account with every funding.

dt = self.datetime.datetime() if not self.position.size: # Entry logic ... else: # Collect funding rate = self.data._dataname.loc[dt].funding if not np.isnan(rate): funding = -self.position.size * rate self.broker.add_cash(funding) # Exit logic ...

Here I am assuming that you are using Pandas DataFeed and there exists a funding column in it with NaNs everywhere except at 0-8-16h. You can pull the historical data using Binance's API endpoint.

@hoflz5 after some more testing I can conclude that this works great. When using 1m data it is very close to the binance live trading stream (but not completely, I guess I would have to use tick data for that).

@ringtail One thing you can try is to check the length once you run the backtest and log it to terminal. This way you only fail once and can run the next test with the appropriate adjustements to start date.

Another thing I do to assist with this problem is to have a start date for data and a start date for trading. I make the room between these two more than enough for whatever strategy I'm running.

Start date for data is the one you are already using. Start date for trading you just put a check at the beginning of next and return if the start date is not met.

I think everything is much easier with travel companies. I have recently booked a trip via http://youtravel.me/. I think you can also check it out.

@csibi-levente14 AFAIU close method will just issue a sell/buy order to nullify (close) whatever position you have - so in case of long position the sell order will be issued, vice versa for short one.

BTW, the issued order is returned by the 'close' method - so you may examine what order was issued.

@tradersnorth said in Forex custom commission and multiplier:

cerebro.broker.setcommission(leverage=50, stocklike=False, commission=comm, mult=100)

AFAIU setcommision method just creates a generic CommInfoBase object using passed parameters to this method.

It means the commission parameter is expected to be (quoting from CommInfoBase docs):

``commission`` (def: ``0.0``): base commission value in percentage or monetary units

In your case, using cerebro.addcommissioninfo method is probably more appropriate. It allows passing a custom CommInfoBase inherited commission object.

I have resolved the issue sorry for the trouble !!.