QuantConnect Community

could you please use the TIME variable to log the time when the new candle was emitted.

and post back if the problem still accours: from the example library a sample with time:

https://github.com/QuantConnect/Lean/blob/c926b68ed452c3afdbbda3eeae2084f53eab859d/Algorithm.Python/UpdateOrderRegressionAlgorithm.py#L118

use this line to test it:

https://github.com/QuantConnect/Lean/blob/cbd953437f987392f2c2e154adcb7d43d684b7f1/Algorithm.Python/DelistingEventsAlgorithm.py#L74

the time on the left site is only the log time......but you want the time when it really happens which tells you the TIME property

oh you are a c# user ok....

in c# its i think even easier...

you will get the idea.....just print what is in Time.ToString()

https://github.com/QuantConnect/Lean/blob/8dc78a3a76944cdece2f9d5a943c4cf1b03eb508/Algorithm.CSharp/BasicTemplateCryptoAlgorithm.cs#L93

but never count on log date....thats what i wanted to say...use algo Time only

I was logging the start/end times from the bars contained within my OnNewHour and OnData methods. I've updated the c# example code with logging that also includes the global Time property.

I'm seeing:

Global Time property is equal to the QuoteBar EndTime property, which is what I expected because I'm emitting the consolidated bar after the previous hour finishes, so all good there.

Global Time property is equal to the OnData Slice bar.Time property, which is NOT what I expected: bar.Time is the start of an OHLC period, so global Time should match the bar.EndTime shouldn't it? Otherwise aren't we peeking into the future?

I have included global Time property values in my log output and attached to my reply.

Extract below:

2018-05-01 01:00:00 Time variable is: 2018-05-01T01:00:00: Start of hour --> 2018-05-01T01:00:00 - 2018-05-01T01:01:00 : o 132.081, c 132.085
2018-05-01 01:01:00 Time variable is 2018-05-01T01:01:00:  Next bar      --> 2018-05-01T01:01:00 - 2018-05-01T01:02:00 : o 132.0815, c 132.081
2018-05-01 01:59:00 Time variable is 2018-05-01T01:59:00:  End of hour   --> 2018-05-01T01:59:00 - 2018-05-01T02:00:00 : o 132.0535, c 132.0535
2018-05-01 02:00:00 Global Time variable says: 2018-05-01T02:00:00
2018-05-01 02:00:00 This is an HOURLY bar from  2018-05-01T01:00:00 to 2018-05-01T02:00:00 : 132.0815 132.082 132.022 132.06
2018-05-01 02:00:00 .
2018-05-01 02:00:00 ..
2018-05-01 02:00:00 ...
2018-05-01 02:00:00 Time variable is: 2018-05-01T02:00:00: Start of hour --> 2018-05-01T02:00:00 - 2018-05-01T02:01:00 : o 132.0565, c 132.06

Starting from the top of the log, I read the following:
1) At backtester time of 01:00 we emit a 1min bar via OnData(Slice data) that spans 01:00 to 01:01, with an open price of 132.081 - this is ignored by the Hourly consolidator
2) At backtester time of 01:01 we emit a 1min bar that spans 01:01 to 01:02 with an open price of 132.0185 - this is used by the hourly consolidator as the 01:00 to 02:00 open price. This is wrong because it's off by one period (in our case 1x one minute bar). If we used second bars, we'd be off by 1x one second bar.

The same problem occurs with the hourly close price - it uses the close of the 02:00 to 02:01 period, instead of the close of the 01:59 to 02:00 period.

My updated logging example (in c#) is attached, as is the backtest.

there is a support tab inside the algorithm lab on the left side if you want to share it with the support team and get an answer

hmm maybe the second bar is used because that is the price of first bar closing

lets what they will answer

Thanks very much for looking at my post and offering some suggestions, much appreciated.