Yes. However, depending on the analysis you are conducting, it might be useful to keep two separate variables, one volume per trade variable for each exchange.

Can you please explain the potential benefit of keeping two seperate variables for each exchange? 

My understanding is that if we combine  data for both the exchanges then it will show the TRUE magnitude of volumes or any data point for that matter. 

Only one exchange is half the picture

It depends on what exactly you want to estimate. In my opinion, if your objective is to capture the representative size of each scrips only, this method would give you a biased number. 

In last decade, the number of trades happening has increased tremendously on NSE (see the attached files from Handbook from RBI). My contention is that we all know the number of scrips listed on BSE is more than 5000 but only approx 2000 on NSE. Despite this number disadvantage, the turnover of NSE is far ahead of BSE. This indicates that there could be substantial difference in the ticket size of individual trades (this needs to be verified from individual scrips trade patterns on each exchange). Another argument is that due to more individual participation on NSE (again needs to be confirmed from type of traders on each exchange), the trade pattern of NSE should be more representative of retail trades. 

What you are proposing is a total value measurement approach which might defy your above expected objective due to this size anomaly of trades on each exchange. If you can recall the cost of capital proposed by Fama-French 3 factor model, they had try to nullify the size effect too.

But if you want to estimate trade size for total market it is unlikely to produce any erroneous result.

Hey Surya,

I agree to your arguments. In today's so called high-end technology backed trading infrastructure measuring the liquidity based on the per trade ticket or trade size is misleading. Partner in crime is high frequency trading. Another aspect is, in today's trading terminal there is an option for the trader to disclose his qty in pre-set multiples.

Then added complexity as you mentioned is more day or swing traders are more active on NSE than BSE causing high turnover on NSE. I feel it can easily be explained by the technology friendly real time data feed provided by NSE, while BSE provides it very expensive that makes BSE more institution friendly. Most trading and charting platforms are ready to run NSE while BSE needs lot of hacking. thats the reason Indian traders are active on NSE than BSE,

Now back to my objective, I feel in some cases Per trade volume has predictive power of accumulation/ distribution. One needs to really master the art of spoting which "per trade volume" is real and which one is manipulated by those high frequency guys. Thats the real art. It can not be only explained by the qunts but really require experience in interprting the whole scenario in that script at that point of time, the information, disclosures, index movement, movement of similar scripts etc.

Taking trade size of total market is baseless and is a fake measurement of overall liquidity. It only misleades.

Dear Ankit,

Agree to your views here. I have the academic view of the issue here. As a real time practitioner, you see this issue with much pragmatism. But given this question, just interested to know, did above formula help you in your objective. Will appreciate your views here.