That's why price feeds should include internal market price of the asset weighted by volume. With growing relative volume of the internal market the external price becomes less relevant.
I'm not sure if current price feed scripts do this though.
In my example internal volume is irrelevant.
By definition the peg "links" the price of a bitAsset (bitA) to the price of an external asset (A). If we account the internal volume (of bitA) in the feeds then the bitA will not be pegged to A.
I disagree. Think of the original concept without feeds. The bitA will be pegged to A simply because market makers make profit by maintaining the peg. As long as the majority has enough faith in the peg it will stabilize itself.
Accounting the internal price would help to gradually switch (with growing volume) to a system without feeds.
Even if we could account for all trades worldwide (BTS <-> A) and devise "perfect" feed the question in OP will still be valid. Shorting bitA is not the same as selling A for BTS.
Lets look into the case where we account the internal volume into the feeds. Imagine the following:
We account for worldwide trade volume of A - $1mil.
We have internal market volume of bitA of $10 mil
In the real world the price of A is X BTS.
In the internal exchange the price of A is X+Y BTS.
With current implementation feed price should be X. In your proposal to account for the internal volume feed should it be ~ X+Y [ derived from (11X + 10Y)/11 ].
Which one looks better for you ?