Learned from the Flash Crash report
SEC and CFTC detailed their findings in the 104-page report on how a single trade overshot the market on May 6, 2010. The trade was done by Overland Park, Kan's Waddell and Reed Financial Inc, a firm with $25B under management. The sell order of 75K contracts of E-mini, which mimics S&P 500, totalled about $4B, was amplified by other algorithm trading firms, including Waddell itself so that the market was precipitated down more than 1000 points in minutes.
The report is in detail on this had happened. From which, we can take a peek on these high frequency traders' algorithms.
Waddell's algorithm scheduled sell all 75K contracts of E-mini at the pace of 9% of volume. That means, the algorithm would monitor what current volume is and then put out 9% of the volume on their sell order. There is no other parameter, according to the report, such as the market has been down 30%. The algorithm would still royally send out 9% of the volume. It is not hard to image that the other HF algorithms would do similar things: if they sniff there is large volume coming, the algorithm would sell, even though decisions seconds ago were buy. So the volume spiked up momentarily. Waddell's algorithm didn't know the price was down and all other algorithms were selling, it still pumped 9% of the total volume. This formed a positive feedback loop. The huge drop occurred.
We can see that these algorithms have different trading parameters. No one except the owners knows what they would do. Had Waddell's algorithm detected price had already down 30%, would it still keep selling? Probably. Here is why:
E-mini is an index used to hedge individual holdings. One case is , when Waddell decided to sell E-mini, they had expected their holdings would rise but in case they were wrong, they sold E-mini to protect them self. If the selling were proved to be correct, i.e., they were wrong and the market down, they would profit from the sell. So the lower E-mini was priced, the more profit they would rake in. To a point lower enough, they needed to cover it back. At the same time, their individual holdings also needed to find another hedge. This ultimately says that they don't care how low price can go before they start to cover. In other words, there is no need to detect current price.
Some algorithms work against this thinking, though. The report says 6 out of 12 largest HF trading firms stepped out trading during this spiral. There is little clue why but we can guess that their algorithms stopped trading. This is like the auto "stop order" got halted somewhat. Again, there is no way to know how the algorithms were programmed.
The report is in detail on this had happened. From which, we can take a peek on these high frequency traders' algorithms.
Waddell's algorithm scheduled sell all 75K contracts of E-mini at the pace of 9% of volume. That means, the algorithm would monitor what current volume is and then put out 9% of the volume on their sell order. There is no other parameter, according to the report, such as the market has been down 30%. The algorithm would still royally send out 9% of the volume. It is not hard to image that the other HF algorithms would do similar things: if they sniff there is large volume coming, the algorithm would sell, even though decisions seconds ago were buy. So the volume spiked up momentarily. Waddell's algorithm didn't know the price was down and all other algorithms were selling, it still pumped 9% of the total volume. This formed a positive feedback loop. The huge drop occurred.
We can see that these algorithms have different trading parameters. No one except the owners knows what they would do. Had Waddell's algorithm detected price had already down 30%, would it still keep selling? Probably. Here is why:
E-mini is an index used to hedge individual holdings. One case is , when Waddell decided to sell E-mini, they had expected their holdings would rise but in case they were wrong, they sold E-mini to protect them self. If the selling were proved to be correct, i.e., they were wrong and the market down, they would profit from the sell. So the lower E-mini was priced, the more profit they would rake in. To a point lower enough, they needed to cover it back. At the same time, their individual holdings also needed to find another hedge. This ultimately says that they don't care how low price can go before they start to cover. In other words, there is no need to detect current price.
Some algorithms work against this thinking, though. The report says 6 out of 12 largest HF trading firms stepped out trading during this spiral. There is little clue why but we can guess that their algorithms stopped trading. This is like the auto "stop order" got halted somewhat. Again, there is no way to know how the algorithms were programmed.
posted by Intl.investment.services at
6:47 PM
0 Comments:
Post a Comment
Subscribe to Post Comments [Atom]
<< Home