Tales From The Hoodie – Or – Are Exchanges Too Big to Fail?



On May 18th 2012 Facebook joined 2,500 other corporations in coming public on the NASDAQ Stock Exchange. The events and chronology of what happened that day, according to the Securities and Exchange Commission, has just been released, with the SEC leveling a $10 million fine against NASDAQ for its role in marring the largest most widely anticipated IPO in US history. This morning’s note digests the SECs facts in the case, and points out key issues.


Although NASDAQ filed for a rule change in 2007 that would remove a “randomization application” that would allow IPOs to start trading immediately upon completion of the IPO Display Only Period (DOP), it had never actually removed that module from the exchange’s matching engine.

Although NASDAQ tested its systems in the weeks prior to the IPO assuming, it had tested assuming 40,000 orders would be entered into the IPO cross (In actuality 496,000 orders were entered by NASDAQ members on the day of the cross).

While NASDAQ also had a “validation check” which would compare the orders entered into the IPO cross with the orders in the exchange matching engine, this check created an infinite loop that prevented the running of the cross. According to the SEC:

The design of the system created the risk that if orders continued to be cancelled during each re-calculation, a repeated cycle of validation checks and re-calculations – known as a “loop” – would occur, preventing NASDAQ’s system from: (i) completing the cross; (ii) reporting the price and volume of the executions in the cross (a report known as the “bulk print”); and (iii) commencing normal secondary market trading.


NASDAQ convened a “Code Blue” conference call to decide what to do. Engineers determined that the validation check was the root of the problem, although they were not sure how (the SVP/INET was not even aware such a validation check existed). The Code Blue conference call members decided that their only option was to override the validation check, despite it creating two known issues:

–          Members’ port connectivity to the exchange would be disconnected briefly, requiring reconnection.

–          The IPO cross would be calculated assuming no cancelled orders in the last sub seconds, and NASDAQ would have to make good on those shares and itself short Facebook stock.


NASDAQ overrode the check and ran the cross at 11:30:09. It assumed it might have a short position because last second cancellations would not have been omitted from the cross. However, it did not realize that the IPO cross application was running 19 minutes behind  due to message traffic and there were substantial cancellations and new orders between 11:10 and 11:30 that were not reflected in the cross. NASDAQ later learned that their short position was 3 million shares.

NASDAQ immediately began receiving calls from members, as the removed validation check and the 19 minute delay also had the additional effect of not properly delivering confirmations of executions and cancellations to their members.

NASDAQ also disseminated stale quotes on its private data feeds, as well as different stale quote on the SIP. The Street was flying blind. Members did not know what their positions were, and they did not have an accurate picture of what the actual order books looked like.

At 11:35 the Code Blue Team talked and decided not to halt trading.

At noon the CEO of a large broker (we all know who that was) sent an email to NASDAQ’s Greifeld asking him if it would be prudent to halt trading since everyone was flying blind. NASDAQ did not respond or change their view about halting the stock.

At 1:50 NASDAQ released more than 13,000 stuck orders (the orders caught in the IPO cross – validation check loop between 11:11 and 11:30). These orders caused Facebook to fall 93 cents in 1 second.

After this release, NASDAQ’s system began to function correctly, and it was then that NASDAQ realized it was short 3 million shares of Facebook, and not the smaller number it originally thought it was short. NASDAQ covered for a $10.8 million profit.

NASDAQ’s malfunctioning IPO cross application also affected trading in Zynga stock – a stock highly correlated to Facebook. Stuck orders resulted from trading halts and unhalts, some of which were in violation of NASDAQ’s halting rules (NASDAQ cut out the DOP).


NASDAQ inadequately tested its systems prior to the largest IPO in U.S. history.

NASDAQ shorted shares against its own rules. This resulted at times in deficient net capital requirements.

NASDAQ violated Rule 4757 by corrupting Price/Time priority.

NASDAQ failed to comply with its own Rule 4120 concerning the DOP in multiple instances.

NASDAQ had discussed the option of halting trading, and decided against it despite known issues and urgings from market participants.

Additional Questions:

How can the SEC insure that stock exchanges, which all have very similar matching engines, put the interests of the market above their profitability? Facebook should have been halted, despite the fact that doing so would have been a) embarrassing for NASDAQ, and b) result in lost profits.

How did the SEC determine that a $10 million fine was appropriate, given the massive losses inflicted on several brokerage firm members – estimates north of $350 million? The fine is actually less than what NASDAQ made shorting the IPO stock. Were NASDAQ’s transgressions only twice as bad as NYSE’s releasing of private data feeds ahead of the SIP (recall NYSE was fined $5 million)?

The sheer scope and compounded errors made by NASDAQ management was …remarkable in this IPO; the punishment could/should have been much higher. This begs the question: Are Stock Exchanges considered by the SEC and the Treasury to be Too Big To Fail? Is the $10 million penalty a slap on the wrist because the SEC is full aware that stock exchange profitability, specifically with regard to cash equities, is currently anemic as it is?

Is the SEC-created web of weak for-public stock exchanges – server farms, strong enough to withstand market forces that might pick winners and losers among them? Can exchanges continue on a path of expanding in all markets and all asset classes, operating for-profit server farms which sell faster speed and access to the highest bidders, regardless of the risks to the public, without free market consequence? Does this small fine create Moral Hazard?  Is Direct Edge too big to fail too? How about BATS? Who decides?

Can stock exchanges be trusted to implement safe systems, with increasingly added complexity and band-aid fixes? Will they be on the hook the next time a market impacting issue arises, such as a LULD failure, a circuit breaker malfunction, or worse?

Will the executives be able to avoid accountability, as this executive has when he denied knowing and understanding the meaning of Gross Profit Margin in a court of law, because they “have accountants who do that for them”?