Workaround for excluding premarket bars when using Barnumber()? Want to get % of trading session elapsed with 1min chart as "(CurrentBarNumber/390);"

[Glefdar]

EDIT 3: The thread I linked below in the second edit had the solution:

#( CUMULATIVE ) COUNT OF THE TOTAL NUMBER OF GREEN BARS (CLOSE>OPEN)
# ON THE ENTIRE CHART WITHIN SPECIFIED TIMEFRAME
# By XeoNoX via Usethinkscript.com
input startTime = 0930;
input endTime = 1600;
def Active = SecondsFromTime(startTime) >= 0 and SecondsTillTime(endTime) >= 0;
def var = close>0;
def cumulative = if Active and !Active[1] then var else if Active then compoundValue(1,  (cumulative[1] + var), 0) else cumulative[1];

Then it was simply a matter of making this replacement for the commented out plot:

#plot scan  = cumulative;

plot barpercent = if active is true then (cumulative/390)*100 else double.nan;

Hi all,

At first I was trying to use SecondsFromTime and SecondsTillTime to do this, but I think it would be less CPU intensive to use BarNumber() for the purposes of calculating what % of the trading session has elapsed on a 1min chart.

For example, if CurrentBar() == 5, then I want that to ideally be the fifth minute into the trading session on a 1min chart. (I realize that using BarNumber() will return false values on highly illiquid stocks that have less than 1 tick occurring per minute, but I think that's an acceptable tradeoff.) Then I want to be able to calculate "CurrentBar()/390" to get the percentage of the trading session (390 minutes) that has elapsed (i.e. how far into the trading day is it, percentage-wise).

However, I can't figure out how to exclude premarket bars without entirely disabling extended hours from appearing on my chart (which I don't want to do). So if there were 15 candles from 4am to 9:29am then BarNumber is returning 16 at 9:30am instead of returning 1.

Is there any workaround? If there was a way to count a running tally of the bars between specified times that would work, for example, but I can't see how that would be possible. Maybe it would be possible in a really convoluted way? For example:

"if secondsfromtime(0400)>0 and secondstilltime(0930)>0 and BarNumber()==15 then MarketOpenBarCount = BarNumber()-15"

But I think that doesn't work because the secondsfromtime and secondstilltime functions only register based on the current time (not making a record of what was conditionally true in hours before the current time)? Since the "seconds" counting functions don't seem to work for historical plotting on intraday charts.

That is, it seems impossible to just plot a line of the seconds that have passed starting from a particular time. For example, this code:

plot data = secondsfromtime(0930);

It plots this (!?):

That's kind of a side tangent but it's the reason why I am trying to use BarNumber() as a workaround for solving this instead of using SecondsFromTime.

Is there any solution for plotting how much time has passed from "X" point or how much time has transpired as a percentage of the total time in a trading session?

EDIT: I think I found something that might give an answer https://usethinkscript.com/threads/consecutive-bar-count-indicator-for-thinkorswim.324/ I'm not sure if this will solve my question, it's hard for me to wrap my head around but I'll play around with it.

EDIT 2: I just found this thread which I think will be more helpful: https://usethinkscript.com/threads/...-bars-on-entire-chart-or-specified-time.5496/

 

[mashume]

Have you found an answer for this yet?

You can reduce the cpu load using your original seconds from/till calculation by adding this:

declare once_per_bar;

at the top of your script.
You can also create a bar counter manually along these lines:

declare lower;
def today_bar_count = if secondsTILLTime(0930) crosses below 0 then 1 else today_bar_count[1] + 1;
plot data = today_bar_count;

Hope that helps,
mashume

 

[Glefdar]

Have you found an answer for this yet?

You can reduce the cpu load using your original seconds from/till calculation by adding this:

declare once_per_bar;

at the top of your script.
You can also create a bar counter manually along these lines:

declare lower;
def today_bar_count = if secondsTILLTime(0930) crosses below 0 then 1 else today_bar_count[1] + 1;
plot data = today_bar_count;

Hope that helps,
mashume

This is quite interesting thanks. I am actually using this as part of a larger study integration. The other components of the study would be affected by using declare once per bar. But that raises the question for me of when the study calculates during the bar. The description of the function is:

"Changes the recalculation mode of a study. By default, last study values are recalculated after each tick. If this declaration is applied, the study is forced to recalculate the last values only once per bar. This declaration can be used to reduce CPU usage for studies which do not need to be recalculated per each tick."

For example, for the first 1min bar of market hours, will the value of the study get calculated (a) at 1 millisecond after 9:30am or (b) does it calculate at 1 millisecond before 9:31am? One interesting advantage of using once per bar is that the bar value will not repaint... From that standpoint maybe it would be ideal if the value of the study calculated (c) at 500 milliseconds after 9:30am so that it is more representative of the value that occurs during the bar period, but I imagine that (a) or (b) is more likely what is happening. Do you know?

 

[mashume]

This is quite interesting thanks. I am actually using this as part of a larger study integration. The other components of the study would be affected by using declare once per bar. But that raises the question for me of when the study calculates during the bar. The description of the function is:

"Changes the recalculation mode of a study. By default, last study values are recalculated after each tick. If this declaration is applied, the study is forced to recalculate the last values only once per bar. This declaration can be used to reduce CPU usage for studies which do not need to be recalculated per each tick."

For example, for the first 1min bar of market hours, will the value of the study get calculated (a) at 1 millisecond after 9:30am or (b) does it calculate at 1 millisecond before 9:31am? One interesting advantage of using once per bar is that the bar value will not repaint... From that standpoint maybe it would be ideal if the value of the study calculated (c) at 500 milliseconds after 9:30am so that it is more representative of the value that occurs during the bar period, but I imagine that (a) or (b) is more likely what is happening. Do you know?

While I don't actually know the answer, I would bet that it is not based on time past the start of a bar but rather on the first tick (trade) of a bar. Remember that bars aren't drawn, even on time charts, if there are no trades that took place. A bar is only drawn when there is a trade. So imagine an illiquid asset which trades in very small volumes. If you're looking at 1 minute candles, there may only be 10 candles drawn for the day (for example look at BRK/A -- 10 is an exaggeration, but the volumes are REALLY SMALL). The calculation would be made after the first trade that falls within a new candle. Don't count on ToS to be time based (ms after event etc...) as it isn't really.

happy trading,
mashume

 

[Glefdar]

While I don't actually know the answer, I would bet that it is not based on time past the start of a bar but rather on the first tick (trade) of a bar. Remember that bars aren't drawn, even on time charts, if there are no trades that took place. A bar is only drawn when there is a trade. So imagine an illiquid asset which trades in very small volumes. If you're looking at 1 minute candles, there may only be 10 candles drawn for the day (for example look at BRK/A -- 10 is an exaggeration, but the volumes are REALLY SMALL). The calculation would be made after the first trade that falls within a new candle. Don't count on ToS to be time based (ms after event etc...) as it isn't really.

happy trading,
mashume

Very interesting, that makes sense. So for a high liquidity stock that has at least 1 tick per second, a study using declare once_per_bar; would paint and stop painting one second after the previous bar closed. Thanks for your insight.