@vishamar said in 1-min data date error:

I have already tried with different timeframes and compressions.

Which clearly indicates the problem has nothing to do with the data feed and associated timeframes and compressions.

@vishamar said in 1-min data date error:

What did I miss?

Showing the code that actually generates the exception.

@fei-xue said in new sell order executed alone with previous BracketOrder(stop loss):

Yes, that is what happened. Is there any mechanism available in BackTrader to support long-only strategy?

You can implement a custom Sizer (Docs - Sizers) and avoid selling if your position is already at 0.

@fei-xue said in new sell order executed alone with previous BracketOrder(stop loss):

If not, I need to manage all active sell order in the implementation of my strategy and cancel all (some) of them before issuing new sell order, right?

This is a MUST, not something you think you might do, but it's a lot better if the platform can think for you. You need to manage your orders and your positions.

It isn't clear what you are trying to say, because not enough information is provided. From your message let's assume the following:

• Bracket Orders are issued. Main Order (Buy) is active and the 2 bracketing (Sell) orders are inactive (they only become active when the main order is executed)
• Main Order (Buy) is executed and the 2 bracketing (Sell) orders become active
• You issue a new Sell order and is executed
• One of the 2 bracketing orders (Sell) is also executed. The other is correspondingly canceled.

Yes, you have a negative position, i.e.: you have gone short.

What is your expectation?

The only thing which seems logical is to canel the bracketing orders before issuing the new sell to avoid them being executed.

You need orders like MIT (aka Market If Touched), but such an order type is not implemented in backtrader. You would couple it with your StopLimit and make them OCO orders.

You would also need to make sure your broker supports bundling both orders together as OCO.

Your real best option is to use intraday data.

@saikee-wong said in How to execute order during gap open on next bar ?:

May be I have misconception on using the backtrader

We don't know, the provided information is not enough to understand what your actual problem is.

@saikee-wong said in How to execute order during gap open on next bar ?:

I issued an order with StopLimit

@saikee-wong said in How to execute order during gap open on next bar ?:

But if there is a gap, it never trigger.

Do you buy? Do you sell? In which direction does the gap go?

@saikee-wong said in How to execute order during gap open on next bar ?:

What I want is, whenever the market break away from the target price, I will enter the market, at that price if possible, otherwise at the market price

If you issue a StopLimit you can only enter at the Limit or at a better price. You CANNOT enter at a worse price (which seems the implication from "breakaway, i.e.: the price runs away from your order) and you CANNOT enter at Market price.

The best you can do is explain your scenario with exact orders, prices and market movement.

You cannot call trade.pnl

• pnl is an attribute of a Trade instance and can be queried but not called
• Trade instances are notified in notify_trade. You can keep a reference to the notified instance

But

• Trade instances are not updated with each price change, they are updated with orders (i.e.: opening a position, increasing/reducing a position, closing a position).

You can track the value of an asset with getvalue (Docs - Broker) and use the difference to the original to assess the profit and loss.

backtrader isn't pandas. The data feeds are a completely different abstraction.

@asjem said in Bulk loading tickers:

@backtrader thanks, let me rephrase the question. Can I load that format (ticker as first column) with QuandlCSV, or do I need to do something bespoke?

If that format has several tickers, which is what you seem to imply in your first comment, the best approach would be to:

• Load the csv with pandas and the method read_csv

• For each ticker you want as a data feed: slice using the the ticker name

• Load the slice with PandasData

  Docs - PandasData

@asjem said in Bulk loading tickers:

It appears to assume one ticker per file

It's not an assumption. It doesn't even assume you have a ticker name in your csv file.

@vishamar said in Cerebro display limit:

When I ran it again for only 3 months

He stated he run the same backtest only limiting the dates. If it is so, that should not play a role.