During the frenzied build-up to the Facebook IPO in recent weeks, emotions overwhelmed good common sense.
It’s called human nature. In spite of the warning signs about a questionable business model and an inexperienced management cadre, the 300+ million Facebook shares offered on Friday 18 May were oversubscribed, by as much as 30 to 1 according to some reports. That’s supply and demand upside down. Institutional investors were reportedly scrounging brokerage trading desks looking for additional shares to bolster their profits.
Mom and Pop were leveraging the baby’s college funds hoping for allocations of stock before the Friday opening. They had missed Google and LinkedIn, and weren’t about to pass up this chance to win the lottery. That’s supply and demand upside down. Everyone was hoping for the “opening pop” that never came.
Many rationalise the lackluster opening to the underwriters’ new-found knowledge in accurately pricing the stock to avoid opening pops. We wish we were that good. No, it’s human nature, it’s supply and demand, it’s just plain greed (opportunism).
Orders came pouring in from all directions. Algorithmic traders were at the ready to capitalise minute by minute, second by second. Institutions that were given early allocations were ready to grab their profits and run. Electronic traders had applied for allocations from E*Trade and had placed their orders.
The Facebook stock price didn't jump. This was no LinkedIn or Google. To the contrary, the stock began to slip
At 11.00am EDT, the flood gates opened and washed away all those dreams. Nasdaq was overwhelmed. The microburst of trading caused the pricing and crossing software to lapse into a race condition. There was no opening, much less a pop. After 32 minutes, by Nasdaq’s reckoning, the pricing was published and the market opened. It was $4 higher than the initial pricing. But by then the queue had backed up so wildly that trade confirmations were stalled and the system collapsed.
If institutional investors cannot get confirmation of their trades, both quantity and price, then their risk exposure is unmanageable. If cancelled orders aren’t confirmed, it doubles that exposure. The algorithmic traders were ready to pounce on any movement, but it didn’t happen. Trading came to a grinding halt. Supply and demand were artificially reversed and the money stopped flowing. The stock price didn’t jump. This was no LinkedIn or Google.
To the contrary, the stock began to slip. The pop pooped. Investors were selling as fast as possible in an attempt to salvage their investment. Underwriters stepped in to slow the decline. The opening pop was gone – and so was the fever that clouded human judgement.
Nasdaq suffered from the same malady of others before it. This is not the time an exchange has been overwhelmed by trading volume – and it won’t be the last. The current global exchange infrastructures are inflexible and cannot scale to meet this kind of demand. The solution is a restructuring of exchange technology architectures and the underlying infrastructure.
The Watson supercomputer won the Jeopardy game last year. The system never failed. It is not an Intel system
Exchanges are facing issues that other industries have dealt with for several decades. The amount of data is growing exponentially, to now literally Zetabytes (1,000,000,000,000,000,000,000). Storage, manipulation and analysis of this much data is a monumental task and not to be taken lightly. Commercial banks deal with these volumes daily, with money flow and credit card clearing. Insurance companies handle extreme volumes of actuarial data daily. Gambling casinos handle these volumes of “trades” every hour.
Maybe the capital markets could take some lessons. These extreme volumes of data are stored on technology designed for that purpose. There is another choice for the exchanges and other trading venues. When all you have is a hammer, everything looks like a nail.
There are only two viable computer architectures left in the high-performance server technology field; Intel x86 from Intel or AMD, and Power from IBM. The Watson supercomputer won the Jeopardy game last year against the best human contestants on the planet. Watson handled billions of pieces of data in less than a second for every question. The system never failed. Watson is not an Intel system. Maybe we put our money on the wrong horse. It’s horses for courses. We need to rethink the future, and choose wisely.
Terry Keene is CEO and president at financial technology consultancy iSys Capital Technologies.