Quote Originally Posted by EB View Post
Mike, the beauty of your forum is that it is full of instant classics. If you can elaborate on the sixth order polynomial expression for volume flows, that would be particularly appreciated. However, if it is proprietary, I understand.

It could also be helpful with respect to the real time 20 day MF because final separation volume to determine LEV/SEV is not known until the EOD. If it could be estimated throughout the day, it might improve the accuracy of the intraday readings.

Also, you mentioned that linear extrapolation was not suitable. This makes sense if the procedure is to take the first 30 minutes of volume and assume that each of the following periods will contain the same volume. However, one thing I've experimented with is to measure the percent deviation of volume for the current day, x minutes into the day, from the average volume for the first x minutes going back y days. Then, apply that percent deviation to the average EOD volume for the prior y days to get an estimate of the current day's EOD volume.
EB

I am attaching an excel spreadsheet used to create the 6th order polynomial. The procedure was to collect a number of days of NASDAQ intraday volume. Then for each day compute the percent of final end of day volume at intraday points. Then average the sets of data to create an average percent of eod volume by clock time. Then a 6th order polynomial was fit to that curve using excel. The spreadsheet should be understandable with this description. Please note that I used California time for the X axis for this analysis. If you were to try to use the same polynomial you would have to convert to Calfirnia time. So if you place a clock time of 7AM (half hour into the trading day) into the x in the 6th order polynomial equations shown in the spreadsheet you would get 16.6% of eod volume estimate out for the y value. So to extrapolate eod volume you divide current volume by y.

Copy of Projected Volume Calculations (2).xlsx