In that case you need to be working with ticks instead of bars.
actually I'm working with ticks.just needed previous spread data to find some starting values, needed for formulas.The procedure I follow is :collect N spread data from ticks, and calculate Mean, weighted_average, Population Deviation etc. based on that N number of ticks.N is initially calculated, based on average number of ticks per day, considering last 10 days or so .(that's D1 Tick Volumes averaged)but, from EA start, until completion of spread collection (receiving Nth tick) , Ticks samples are too small* to be reliable for Mean and Min and Mode,and DEV calculations...SO, you have any other idea how to achieve a reliable-enough starting point to answer this question :What range of spread is considered normal for the current symbol ? (considering this question is asked right after EA start) [ normal being defined as a range in population of data, that %80 or %90 or so of values happen in that range ]hope I make sense, please share your idea if you think it could improve the procedure.ThanksEDIT:
* : not after a while. I use Sample-Standard_Dev formula when 10% of N ticks are arrived already. but that also needs a good starting MEAN or MODE value , to be valid