Bitcoin Forum

Economy => Trading Discussion => Topic started by: Eric14 on January 19, 2014, 09:27:51 PM



Title: Computing trading open and closes
Post by: Eric14 on January 19, 2014, 09:27:51 PM
I've been wondering what is the standard method for determining and open and close price for a time interval?

Here's two methods I've been investigating:

1.  Interpolate around both start and stop points to get estimated price at exact time intervals.
2.  Take first and last trades inside the time interval as open and close

The first method can be slower because you need to bracket the intervals and sometimes trading can be slow so it takes a while to determine the close price.  However this seems like a more realistic way to estimate the end point of the price during that time interval.

The second just shifts trade values in time (which is faster) to estimate the open and close.  The problem here (and I've seen this on some internet charting tools) is that the intervals might not match up between close and open of the next interval.


Title: Re: Computing trading open and closes
Post by: Rannasha on January 19, 2014, 10:50:18 PM
It is not necessary for the close of the previous time-period to match the open of the next one. In regular stock markets this is even more obvious as the markets are only open for part of the day and developments after market close can make the next days open considerably different than the previous days close.

With that in mind, the second method seems to be the go-to choice. It's easier, faster and more natural.


Title: Re: Computing trading open and closes
Post by: Eric14 on January 20, 2014, 03:55:32 AM
In regular stock markets this is even more obvious as the markets are only open for part of the day and developments after market close can make the next days open considerably different than the previous days close.
I think this is different than interval to interval close-opens.  The trades that accumulate when the market is closed is dealt with the opening cross.  However I believe candles are supposed to line up with either the top or bottom depending on the trend interval to interval.

It seems to me method 1 would be more accurate and a time average seems more natural to me than discrete jumps.