@backtrader said in How to get the next trading day's date?:
That won't give you the next trading day. It only peeks into data (when preloaded) in the future, which is of course cheating and not portable to non-preloaded data. Be your own guest.
Not peeking into data. Peeking into calendar on data.
@run-out
Cool! Exactly what I want. Thank you so much.
I mean "the next trading day" according to the data line, which is a line of daily bars. For example, suppose 4/30/2018 is a trading-halt day and no data on that day, and then 4/27 should be the end of month, but calendar module thinks 4/30 is.
Another use case is there may be data point missing on a particular day, say 4/30/2018, and then I want to use 4/27 as the end of month.
It would be too much work to add every holiday or data-missing days to the calendar. So the goal is to get dates from data lines, whatever they are. Does there exist something like self.datetime can do that? Thanks.
Hi,
My strategy is supposed to deduct monthly broker fee at the end of month. I think it could be done by comparing next trading day's month number with today, but what's the proper way to get the next trading day's date? Or, is there a convenient way to determine month end or year end? Thanks.
@vbs Have you stopped developing backtrader_plotting? The interactive plots are very nice and helpful. Can you please fix the plotmaster issue? Thanks!
@backtrader said in Please help to understand cash changes:
Which obviously has no times 10 (i.e.: the multiplier in the above calculation).
In the first post, I put the code
comm_rb = CommInfoFutures(commission=1e-4, margin=0.09, mult=10.0)
I use 0.9 because 0.9 = 0.09 * 10.
@backtrader said in Please help to understand cash changes:
Cash remaining in your account after you enter the market: 20000000.00 - 2615 * 0.9 = 19997646.50
Closing price: 2609
Cash to be added to your account at the end of the day: (2615 - 2609) * 10 = 60.00
Cash at the end of the day: 19997646.50 + 60.00 = 19997706.50.
The margin cost at the end of day should be based on close price 2609 instead of the price when entering the market 2615, so cash at the end of day should be 20000000.00 - 2609 * 0.9 + 60, instead of 20000000.00 - 2615 * 0.9 + 60.
About the value. My understanding is some money could move between margin cost and cash in the account, but the total account value should not be affected by margin cost. I think the account value at the end of day should be 20000000.00 + 60, instead of 20000000.00 + (2609 - 2615) * 0.9 + 60.
Sorry if the calculations confused you. I don't think more code helps since the problem and reasoning are very simple. More code just bring more confuse.
@backtrader said in Please help to understand cash changes:
I would go into the market and buy 1-future of whatever and see how things work. Real life experience does really mean something.
Yes I did.
Thank you for looking into this.
The order type is stop with trigger price at 2615 and trade fill price is also 2615 without slippage. The OHLC prices are below:
I think I know what's going on. Below is the output with commission set to 0. The margin cost based on fill price is 2615*0.9=2353.5, and the margin cost based on close price is 2609*0.9=2348.1, so the difference is 5.4. The method notify_cashvalue() calculates cash and value based on the margin cost 2353.5 at the trade fill price 2615, but calculates the daily profit based on the end of day close price 2609, which is (2615-2609) * 10 = 60, so the value at the end of 3/19 is (20e6 - 2615 * 0.9) + 60 + 2609 * .9 = 20000054.6, and the cash is (20e6 - 2615 * 0.9) + 60 = 19997706.5. Both match the output.
I'm not familiar the calculation detail at a real broker. Intuitively, the account value calculation should not involve margin cost, and the end-of-day cash should be based on the close price instead of fill price. Maybe the method notify_cashvalue() is supposed to return the cash and value at the moment of trade?
2018-03-16 - cash = 20000000.0 value = 20000000.0 value-cash = 0.0 close_price = 2689.0 margin_val = 2420.1
2018-03-16 - Close = 2689.0 Position size = 0 order = None
2018-03-16 - Entry Short: Order status=1 ref=1 price=2615.0 trig_price=2615.0
2018-03-19 - Order status: ref = 1, action = SELL, Price = 2615.0 Size = -1 Cost = 2353.5 Comm = 0.0
2018-03-19 - Trade: ref = 1 price = 2615.0 value = 2353.5
2018-03-19 - cash = 19997706.5 value = 20000054.6 value-cash = 2348.10000000149 close_price = 2609.0 margin_val = 2348.1
2018-03-19 - Close = 2609.0 Position size = -1 order = None
2018-03-20 - cash = 19997726.5 value = 20000072.8 value-cash = 2346.300000000745 close_price = 2607.0 margin_val = 2346.2999999999997
2018-03-20 - Close = 2607.0 Position size = -1 order = None
Hi,
Is it possible to have some indicators in event-based mode and others in vectorized mode in the same strategy? For example, in a dynamic strategy using _nextforce = True to make it event-based, but keep other SMA indicators in vectorized mode. It looks to me that once the dynamic strategy is forced into event-based, all other indicators are in event-based too.
Thank you.
Of course not if you are not interested.
I've explored the doc you listed but no clue. The custom commission scheme is in the top post. Appreciate any insight but that's ok if not. Thanks anyway.
@backtrader said in How to plot two indicators on the same subplot?:
Docs - Plotting - Option:
plotmaster
Cool. This works!
Just a suggestion. Backtrader plots are based on matplotlib, while it may be better to be interactive plots for analyzing, such as plot.ly or bokeh. There is a 3rd-party add-on based on bokeh https://github.com/verybadsoldier/backtrader_plotting , but it's kind of buggy (at least for plotmaster). In the long term, maybe backtrader can consider migrate to bokeh or other interactive plotting library?