3 ms·
If my investment strategy is "buy stocks and pray that some random HFT fuckup in the next 40 years doesn't wipe out my entire life's savings along with everyone
by joezydeco 13y ago
If my investment strategy is "buy stocks and pray that some random HFT fuckup in the next 40 years doesn't wipe out my entire life's savings along with everyone else's", then the evidence is growing year by year that this strategy is going to be a losing one.
- minimax 13y agoYou must have missed the part where the NASDAQ bug caused about $60MM in losses whereas the subsequent price decline has destroyed $40B in market capitalization. You can lose your shirt in the stock market, but that's been true since long before we started using computers to trade equities.
- joezydeco 13y agoNo, I fully caught that part. I also understood why the NASDAQ computers couldn't finish the cross: because HFT computers were hammering the NASDAQ with cancel orders. You don't DDOS a server and say "oh, the server has a bug." And, to be perfectly honest, I have no sympathy for anyone that lost money on the FB IPO. For once a company executed an IPO and got every cent on the table. And I know that just infuriates every suit on Wall Street to no end, which couldn't make me any happier.
- mooktrader 13y agoAlthough HFT and other algorithmic traders are certainly prevalent in the markets and probably in this IPO, I think this flaw was elicited because of the unprecedented and overwhelming amount of non-professional human order flow. There was a ton of buzz and participation in this IPO. The uncertainty around the opening process compounded the problem.
- Dylan16807 13y agoThere was a bug. A bug serious enough that you could trigger it with low frequency trades. Every time the cross algorithm restarted it did so using data that was 99% stale. This meant it didn't just need to find a few milliseconds of stability, it had to wait until the average order rate went low enough and stayed there for minutes. A relative handful of people changing their orders a few times a minute would be enough to keep such an O(n^2) algorithm occupied.