@pc is correct. Market Pegged Assets against a thinly-traded token are a recipe for disaster and abuse. I would not recommend having this price feed managed by the elected Bitshares witnesses.

Quote from: crypto123 on August 09, 2016, 12:00:26 am

Ok, so only those shorting the Gridcoin MPA would potentially be affected, not those holding the MPA (because they'd be saved by forced settlement) ?

In the scenario I described, yes, for some definition of "only".

But an attack in the opposite direction is also possible:

* Attacker short-sells a lot of bitGRC near the feed price.
* Then, he dumps the GRC price to (for example) 10% of its normal value.
* He creates another (small) short with minimum collateral, i. e. about 20% of the normal GRC price.
* Then, he pumps GRC to >20% of its normal value. This will trigger a black swan, and bitGRC holders are forced to settle at about 20% of the normal price.

I'm not talking about settlement.

Suppose that the feed price is derived from external exchanges, that the total trade volume on these exchanges is low (as seems to be the case right now).
Suppose that external exchanges have not only low volume but also little liquidity and possibly a high spread (I haven't checked this). That means it is possible to move the price by a large amount with only little trade volume.

* An attacker with sufficient BTC in their pockets would try to buy up as many bitGRC on OpenLedger as he can get, at a price slightly above the feed. This may take a couple of days.
* Immediately before the attack, the attacker examines the short positions and their respective collateral. For each short position, he computes the black swan price.
* To execute the attack, the attacker positions a sell order for each short near the black swan price.
* Then he starts "manipulating" the price on the external exchanges.
* As the price of GRC goes up and takes the feed with it, the short positions will be margin called and will buy into the existing sell orders, consuming (almost) their entire collateral.

The attack is profitable if the cost for price manipulation is low in comparison to the value of the bitGRC the attacker has amassed.

The price manipulation attack has been considered before. AFAIR bm's response was that the attacker will end up with BTS that's worth less than what the attack costs. This argument is valid when you try to manipulate the price of BTS against FIAT, or against a high-volume currency like BTC. However, I think with a low-volume asset the attack can be profitable, in particular because the price of BTS itself is not damaged by the attack.