You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
HFT orders revealed
- Thread starter fhl
- Start date
I wonder if this is an attempt at manipulation. The robot creates a sense of great activity by placing many limit sell orders above the market, then lowers the bid. Others might think something is happening and sell. A short term downward price change might follow.
The price change might be small in most cases but sometimes a larger price break may follow. Perhaps it's that bigger price break that the robot is intended to trade.
I am reminded of the Jesse Livermore story about skinning skinners, the bucket shop operators, from Reminiscences Of A Stock Operator, page 54 in my copy of the book. In this story, Mr. Livermore places limit buy orders under the market price, then manipulates the price lower. When the price values rally Mr. Livermore sells at a profit. Maybe the picture below records a millisecond version of the same idea.
<img src=http://www.elitetrader.com/vb/attachment.php?s=&postid=2916005 \img>
The price change might be small in most cases but sometimes a larger price break may follow. Perhaps it's that bigger price break that the robot is intended to trade.
I am reminded of the Jesse Livermore story about skinning skinners, the bucket shop operators, from Reminiscences Of A Stock Operator, page 54 in my copy of the book. In this story, Mr. Livermore places limit buy orders under the market price, then manipulates the price lower. When the price values rally Mr. Livermore sells at a profit. Maybe the picture below records a millisecond version of the same idea.
<img src=http://www.elitetrader.com/vb/attachment.php?s=&postid=2916005 \img>
Attachments
Quote from heech:
Frankly, any HFT unable to filter out insignificant quotes at o(1) cost deserve the ass-whooping they must receive daily.
Yes, so the theory that people are doing this to slow competitors down seems to be hogwash. More likely is that two or more algorithms are interacting with each other in bizarre and unpredictable patterns.
that would kind of defeat the purpose, don't you think? if you're going to try overloading competition doing this type of thing, i'd think you'd be smart enough to write something preventing being overloaded yourself.Quote from sprstpd:
They would have to process their own quotes too.
if you were aware something was up. of course, cat's out of the bag now.I bet filtering them would take about as much time as just doing some numerical calculations with them.
Quote from propseeker:
that would kind of defeat the purpose, don't you think? if you're going to try overloading competition doing this type of thing, i'd think you'd be smart enough to write something preventing being overloaded yourself.
But that is precisely the problem. If you create quotes with lots of orders and you rely on a quote provider for your quotes, you are going to be receiving the same quotes you are dumping to the market. So you will have to filter them too. This theory seems ridiculous to me.
no, if you know when and where a quote flood is coming, you can easily avoid processing them.
i don't know about you, but the architecture we run can turn off parsing at the symbol level. cusip is usually one of the first bytes in a msg. you shave a non-trivial number of cycles ignoring, especially since most of the heavy lifting comes afterwards. 5k updates per second x 300 symbols is 1.5 million updates per second that a competitor is running through his entire engine that i'm not. not so far fetched when you think about it in those terms.