Wednesday, March 28, 2012

Financial Sabotage: How BATS was Brought Down


Yahoo! News: BATS Global Markets: The 'spectacularly botched' IPO:
Friday was supposed to be a coming-out party for BATS Global Markets, the flashy new stock exchange that recently became the country's third-largest trading platform behind the New York Stock Exchange (NYSE) and Nasdaq. BATS was scheduled to launch a lucrative IPO that would cement its status as one of the market's big boys. One computer glitch later, BATS' share price plummeted to a few pennies, and the company's IPO lay in ruins.
BATS (Better Alternative Trading System) Global Markets is based in Lenexa, KS, with offices in New York and London. Unlike their old-school competitors who still use trade floors, BATS executes all trades electronically.

Last Friday, BATS learned the limitations of a lightning-fast, highly-integrated computerized stock market. While the Yahoo! News story linked above places the blame on a "computer glitch" the truth may be that BATS was sabotaged by one of their competitors.

zerohedge: SkyNet Wars: How A Nasdaq Algo Destroyed BATS:
What happened is that a malicious, 100% intentional Nasdaq algorithm purposefully brought BATS stock to a price of 0.00 within 900 millisecond of the company's break for trading! This is open SkyNet warfare. 

The fact that the BATS exchange itself halted just prior to break only facilitated this (and could potentially be a case of malicious sabotage). But one thing is clear - as the data below shows, there is no doubt that an Intermarket Sweep Order originating on the Nasdaq exchange was unleashed to make a mockery out of BATS. It succeeded, and in doing so may have destroyed not only BATS chances for going public, but ultimately ruined the firm's credibility. Who would stand to gain from this? Why exchanges such as Nasdaq and NYSE of course, which already are scrambling for revenue, and in the aftermath of the failed Deutsche Boerse merger, it means that any dirty trick in the book to extend and pretend is now fair game. Such as the algo that crashed BATS.
To understand how a rogue trading algorithm destroyed the BATS IPO in less than one second, you first have to understand that an equity's price is driven in part by bids to buy/sell that equity and the incredibly fast withdrawal of those bids. That's what's illustrated in the graph above. Both the blue and pink lines illustrate how the price of BATS was walked off a cliff in nine tenths of a second. The blue line is on a logarithmic scale making the pattern a little easier to see. The data behind that chart is in the excel spreadsheet below.

From the graph above it looks like every 25ms or so the NASDAQ initiated algorithm would cancel its pending bids for BATS and issue another bevy of bids at a lower price. Because the market maker is a computer which thinks that their's significant demand for the equity at one price it, apparently, is easily enticed to think that that demand has simply moved to a lower price. At that point the equity's quote price falls and the process of ratcheting the price down further repeats.

The use of many simultaneous trades is sometimes called HFT (High-Frequency Trading).

Nanex has more about this problem in an article titled: HFT is Insatiable - its Hidden Costs:
We have spent the last 24 years working with real-time market data on a tick-by-tick basis. We monitor our commercial datafeed in real-time to stay on top of market changes or issues. This past year, we have spent considerable time and effort studying the relentless growth of equity quotes. Based on our findings, virtually all of the additional quotes contribute zero or negative economic value to stock pricing, because they are either way outside the market or end up expiring before any investor or trader could possibly act on them. Furthermore, we can't find any self-limiting mechanism in place that will ever put a stop to this unnecessary and expensive growth of misinformation. The only thing that prevents a sudden explosion in quote traffic is the capacity limitation set by SIAC which runs the Consolidated Quote System (CQS) for the exchanges.
Nanex has also has "a proposal for improving the market for everyone" which opens with this executive summary
In 2011, exchange fees for CQS were approximately $500 million. Per Reg NMS, exchange fees are apportioned to each exchange depending on how often an exchange's quote is at the NBBO. Reg NMS also describes quotes that change more than once per second as flickering quotes. Per Reg NMS, flickering quotes are ineligible to set the NBBO, and exchanges are free to ignore them for the purpose of trade-through protection. We believe that by formally recognizing flickering quotes for what they are, most of the latency games will disappear. 
Therefore we propose to:
  • Set the quote condition for flickering quotes to a new condition called Immediate. Traders that want to enter a quote that can be canceled in less than 1 second, simply need to indicate this by marking the quote with the Immediate condition.
  • Require regular and auto-exec quotes to remain active (and not canceled) until executed or 1 second elapses. These quotes remain eligible to set the NBBO.
  • Require the time stamp field in all quote messages to be set to the time the quote was created at each exchange, instead of when it is transmitted by the network processors (SIP). This will allow all downstream users to accurately compute the age of each quote.
These simple changes will provide an incentive for exchanges to attract liquidity providers who are willing to leave their quote in the system for at least one second, because only those quotes are eligible to set the NBBO which determines the percentage of exchange fee revenue. Also, this proposal will give higher visibility to orders from investors who truly wish to buy or sell a stock. And finally, the changes to the time stamp will allow all participants to determine the age of each quote, and therefore any delays in the system.
That seems like a reasonable first step because, as it stands right now, the stock market really isn't a safe place to be if some rogue trader with an algorithm can flash crash a stock in less than a second.



No comments: