Nanex Research

Nanex ~ 12-Feb-2014 ~ Bad HFT Algo Blasts 2.5 Million Bogus Orders


On February 12, 2014, starting at 13:31 and running right up to market close, a high frequency trading (HFT) algo placed and canceled 5,000 to 30,000 orders per second in individual stocks and ETFs. In many 1 second periods, there were more than 20,000 bogus orders in a symbol, which means each order lasted 50 microseconds (millionths of a second) or less. In 50 microseconds, information can only travel about 9.5 miles (15.2km), which means each of these orders were already canceled after traveling less than 9.5 miles outside of the exchange data center in New York (New Jersey). Worse, only computers half of that distance could have physically acted on that information fast enough. In effect:
anyone farther than about 4 miles away, was essentially processing 20,000 bogus "real-time" quotes per second in 1 stock.
There were 30 symbols affected: all beginning with the letter 'O', 'P', 'Q', 'S' or 'T'. Not all stocks were listed on the same exchange, but as far as we can tell, these blasts of extreme quotes all had Nasdaq as the reporting exchange, which is interesting in light of the recent statement Nasdaq issued about the cause of the August 2013 blackout (more than 26,000 quotes per stock per second).

Very few trades executed during the time when a symbol was bombarded with high order cancellations.
In one 15 second period of time, there were 275 thousand bogus orders placed and canceled in 1 stock: PLUG. In the nearly 8,000 other NMS symbols, there just 255 thousand quotes combined. Which means for 15 seconds, one high frequency trader accounted for more than half of all stock market quotes: all from orders they placed and canceled in one stock.
We documented a similar HFT quote spamming algo that rocked the market on November 2, 2012.

Below is a list of symbols affected along with the number of quotes and count of HFT quote blasting seconds.

Here is a PDF of the individual events. Note how many occur in the last 3 minutes of trading.


Nanex Research

Inquiries: pr@nanex.net