@josephjospeh I run the full test from the strategy start, but it is traded less than a year. Helps me also to keep comparison between test equity and real equity to understand if test was made correctly.

@josephjospeh said in Trade-along with backtrader:

I was just imagining a webapp that will display new orders to be manually inputted,

Print corresponding message while issuing order within the next()

display historical trades with some charting app (hopefully tradingview) for analysis and

bt can do this by standard means

portfolio analysis with pyfolio. Good luck to me then.

bt can do this with older version of pyfolio

For "pause > append data > resume mode" you will need to come up with some kind of live data feed. It is possible but depends on how low-tech you are. I just re-run the test. It is much simpler and can be done much faster compare to developing all bells and whistles.

Code the strategy, run the script every day after market close, print size, ticker and direction for orders bt issued after current bar, manually put these orders to your broker account next morning. Next time you run your bt code, it will execute orders and you will have synced bt and real broker accounts.

As a result you should have backtest performed in parallel with real trading.

I doubt that you need Django for this, but it is your choice to add more complexity than it requires. :)

Check out optimization part here - Docs - Let's optimize

I didn't make changes.

@jf not sure what to answer. I just wrote a code posted above.

Within the strategy I defined self.sl_price variable which kept stop price calculated for every open position. This variable can be obtained from observer using self._owner.sl_price name.

Backtrader executes market orders on the next bar open by default, without any cheats. But no next bar open price legitimately available in this case since it is the information from the future.

Just put together the script, run it and print time of every next() delivered. And you will know for sure, probably quicker than get your question answered.

@alexander-lifanov seems the mistake is in your code.