Yea, good point, it all depends on where the volume was. I would break it up this way: The initial sell off: These orders would be executed from traders that u

Yea, good point, it all depends on where the volume was. I would break it up this way: The initial sell off: These orders would be executed from traders that use a live twitter feed to make buying or selling decisions. The underlying algorithm would have been able to execute these orders faster than any human. The serious news would have driven them to sell quickly, causing the market to start diving. Most of these probably made money, depending on how they were able to react to the news that it was fake. It is possible that there were algorithms selling all the way down the entire spike with the expectation that it was going to go down even further, losing money on most of the positions during the bounce. After-all, the bounce was faster than the drop. Secondary sell off: While the algorithms are executing sell orders, market makers would begin to realize that a huge volatility spike was entering the market place, whether they heard the bad news or not. Because market makers train their models on relatively low volatile intra-day data, this spike would have been enough to reach either max positions or max downdraw, forcing the traders (or the market making algos they created) to cover positions for a loss on the way down.