@sagor: The effective volume indicator in the download section has a small bug in the formula and does not show cumulated effective volume as used by Pascal Willain. Also there is no differentiation between large and small effective volume.
To build a correctly working effective volume indicator, the following steps would be necessary:
(1) load a secondary bar series of 1 minute data, which is used by Pascal Willain to calculate effective volume
(2) load the effective volume values into an array to determine the median, or alternatively use a moving average to divide effective volume into small and large effective volume
(3) calculate cumulated large effective volume and cumulated small effective volume and plot it on a chart
I think that it would be worth the effort coding the indicator, as the same approach can be used for similar indicators such as On Balance Volume or the Williams Accumulation Distribution indicator, which belong to the same family as Effective Volume.
The following 2 users say Thank You to Fat Tails for this post:
I'm unable to open the code (can this be done outside of NT?). In any event reading the code of the 6.5 inidcator would mot likley confuse me. However I have read Pascal's Value in Time and have a fair idea of his approach and have been thinking about how the indicator might be emplyed for intra-day trading. He uses 1 minute bar to arrive at an EV for the Day (405 intervals). For intraday trading (say a five minute bar) we would need to calculate EV in mili second periods (.666 of a second) and then rank them for each five minute bar, find the mode (?) and then seperate them on that basis into large and small EV per bar. Then cumulate the results. Would such a small time increment be necessary?...Perhaps that's over kill and every five or ten seconds might be satisfactory. Would it be computationally demanding?
I've thought also it might also be useful to look at delta volume as the input for volume, or have the alternative to do so in the indicator.
It's worth noting that the first minutes volume (and possibly the last imo) are ignored in the calculation.
I have some code MQ4 code I found on the i-net a while ago with a view to look at develping soemhting for NTrader. Not sure if it' of use for you guys. I will try to post it here.
@Bondi9999: The last MetaTrader indicator seems to do it correctly. It takes the effective volume over a rolling period of 24 hours and then calculates the median. This way you get a threshold to make a distinction between large and small effective volume.
There are a number of problems, that would need to be solved for application to futures:
- Regular session / night session: If you use the volume average over the entire session, probably all night session volume bars will be classified as small volume and the larger line of cumulated effective volume will not move. This would be a problem for FOREX futures, as the night session activity will not be correctly represented.
- Holiday sessions: After a holiday session the median calculated from effective volume will come out too low. Nearly all volume will be classified as large volume.
This shows that the approach of a 24 hour rolling period is not sufficient, but that a longer period, maybe two weeks should be use (the period could be user selectable).
For the night session problem I do not have a solution. Pascal Willain uses the method on stocks, and there is no night session.
The following user says Thank You to Fat Tails for this post:
Fat Tails - What about converting the EV outcomes to a ratio of the total volume or total delta volume (or whatever the input) from which it was calculated for each incremental period, then the variation in volume between periods you are referring to would be irrelevant?
I think for the indicator that having this as an alternative would be useful.
In summary I'm thinking as user inputs:
(1) The choice of the time frame to calculate the individual EV inputs (For example, for trading off Daily charts Willain uses 1 minute EV increments, calculates daily EV small and large and then uses them as cumulative. For trading intra-day trading one might want to use 1 secondor 5 second EV increments (or something else..not sure at what point it is meaningless), calculate EV for the 5 minute bar and then use them for the defined cumulative period.
(2) Choice of the volume input, either pure volume or delta volume(guess the latter would use gomi's).
(3) Choice to use the ratio approach or the raw volume /delta volume approach to calculate EV increments.
(4) Choice to (a) exclude the first minute of the day and (b) the last minute of the day...(thinking about it probably should leave the last minute in given the importance to portfolio valuation of the close and the clue to institutional positions.. however to have the alternative would be nice).
(5) Choice of number of periods calculate the cumulative over. This is useful as one might want to always calculate a rolling five day summation for example, or might want to calculate the cumulative from a last major swing point or large volume bar.
(6) Choice to plot (a) Total EV, (b) EV Small, (c) Large EV..All or any combination
(7) The choice to also plot one or two an averages of the cumulative result and the choice to simply plot those on their own or with the EV's plotted also.
(8) The choice to plot as a histo gram.
The resulting information would be employed similarly as delta volume however imo it may be more valuable because a bit like Force Index it brings to the table the effect of volume on price and therein some clue of what the insto's are trying to do. Willain uses it with Active Boundaries. There are better tools we have than that imo for day trading to work out where we are contextually, however whatever one uses I think EV can add value.
The only other considerations I can think of are, firstly, how would one work with other bar types than minute? I would imagine one would use the multi-time frame functionality or create some type of user symbol for the EV and then break down the EV calculations based on the type of bar being employed. And, secondly, system trader may want some sort of out put. I would imagine there are so many variations that might be employed it would be best to leave that to individuals who no doubt can code anyway.
I'm happy to have a go at all this however it will take me an age as I will need to learn C# from the beginning
P.S. As I was reviewing my post, the thought occurred that what would be very interesting to see would be VWAP using the volume amount that is regarded as large by Willain as the input. (Using the VWAP that enables day and specific time to calculate from...as I like to see VWAP from turning points as well as on the day).