This is the one piece of the puzzle that took me the longest to come up with a logical explanation for. After MtGox disabled BTC withdrawals, I expected the premium over BitStamp to disappear PLUS a bit more to reflect the fact that limited fiat withdrawals in yen and euros were still possible. But if I had fiat on Gox, I would have deployed all of it when the price premium was 20-30% below BitStamp. I simply don't buy the "it's easier to sue in USD" to explain the price dropping under $100!
My explanation for the dumping down below $100 follows from my theory (see links below) that Gox had been running a fractional reserve in bitcoin since 500,000 - 1,000,000 BTC were stollen in 2011:
https://bitcointalksearch.org/topic/m.5397869
https://bitcointalksearch.org/topic/m.5438572
The theory states that Gox was using client fiat funds to purchase bitcoins from other exchanges and individuals in order to keep up with bitcoin withdrawals. Prior to disabling bitcoin withdrawals, they had, say, $150,000,000 in client funds sitting in USD IOUs but only $30,000,000 in actual cash. If this fact was revealed during bankruptcy, it would have been clear-cut fraud and Mark would be doing jail time: how could you explain $120,000,000 in missing cash??
So, when Mark knew it was over, he began selling fractional reserve GoxBTC from multiple alias accounts (to make it look like real customers were selling). He drove the price so low, that most people converted their USD funds into GoxBTC. He kept driving the price lower and lower hoping that enough people would convert into GoxBTC so that he could rebalance the USD funds owed to customers with what was actually in the Gox corporate bank account. He drove the price so low that people actually wired additional funds to Gox further reducing his USD solvency gap.
But it looks like it wasn't enough. I think the bankruptcy filling declares $50,000,000 in fiat liabilities and $30,000,000 in cash. But my point is that if he had gone bankrupt prior to dumping all those bitcoins (that didn't really exist), the fiat deposits owed to real customers would have looked ridiculous and screamed "fraud."
In conclusion, the price was driven low by Mark dumping GoxBTC to entice GoxUSD holders to transfer from USD into GoxBTC, in an attempt to close the USD solvency gap between what Gox had in its bank with what it owed to its clients.
Thanks for this thoughtful post. I'd wondered about it too and never bought the USD recovery idea either. However, how does your explanation account for the huge volumes on Gox? If Gox was making .6% commission they made almost 180K coins their last month of operation according to bitcoin charts.
How are you calculating such a number? Even assuming the "most favorable conditions", its just 6000.
That being, 1 million exchanged * 0.006 = 6000
Really though, assume maybe a 0.004 fee average, due to tiered fee rates, and of course that is 4000.
In the 3 months before, They got about 2.3m volume. So another 9200.
Assume 50% fee fiat -> BTC, all at $136: Another 50,000BTC. However, we know this didnt happen(they didnt acquire 50kBTC @ 136). So, could we say.. in the last 4 months, they covered 65000/744000 of liabilities, assuming $136 exchange rate?
Otherwise... if we go by the BitStamp exchange rate, then they acquired 12,500 "real BTC" in fiat fees, and another 14,000 in BTC fees.
Total in 4 months: 26,500 / 744,000 = 3.56%
--
Now, honestly, thats not too bad. But, even assuming just a 20% increase in BTC value per year, over the 10 years that it would take them to pay out "BTC:BTC", the BTC liabilities value in USD would average out to have quintupled.
What would be interesting is a real timeline of liabilities. I'm curious what the financial state of Mtgox was before this "hack" in early February. Were we still on a sinking ship or was Mtgox solvent?