How often do traders know news before you release it?
I was in the car listening to a business program on satellite radio, and they were talking about crazy moves in shares ahead of news. The host, a trader and money manager, said, “Even with Reg FD and all the disclosure rules, I swear 85% of the time there’s leakage.”
In this age of mandatory dissertation in television ads of pharmaceutical side-effects, one approaches the term “leakage” with caution. But in market-structure, the behavior of money behind price and volume today, leakage is a functional fact. Routinely a day before important news, high-frequency trading jumps, signaling impending change and oftentimes distorting price ahead of material information.
Recently a client had a secondary offering for a big holder. Two weeks before it, the broker who would later underwrite the offering led in what we call influential order flow. Two other brokers had matching increases. Yet the company hadn’t yet determined a manager.
Both examples imply leakage. Maybe the fast traders picked up a nugget of information? Word worked around to broker trading desks? Both are possible. But we doubt shenanigans played a role. The tipster in both these instances and in much of what appears predictive in stock-moves is the same: Math.
High-frequency traders, the ones signaling money-moves – the reason ignoring fast trading the way so many do is like yanking the fuel gauge from your car and hucking it in the trash – aren’t people. It’s not somebody on phone calls or in meetings. No channel-checks have been made or hedge funds plied for leads. They’re machines programmed to respond to data.
Therefore, Watson, the elementary conclusion is that the math changes before news. Now could somebody in the know be the trigger? Sure. Machines seek central tendencies and departures from standard deviation (as do our algorithms) and a trader unaware could unwittingly tip them to a directional shift with orders that don’t look like the others.
The machines spit feeler orders like tiny drones. For example, a utility with market cap around $2 billion trades 225,000 shares daily. Over a third of the roughly 2,000 trades yesterday were for less than 100 shares. Ten percent were 10 shares or fewer. Once we saw six hundred-share trades move CSCO $6 in one second. High-speed trading algorithms feather the markets with the smallest possible commitments seeking directional tips.
Suppose a computer metered human traffic at New York’s Penn Station. The computer would know that before the 2:07 to Bay Head on a westbound track, people are going to fill the station. Now imagine the computer can determine how many cars should comprise the train, using algorithms that measure human traffic the half-hour leading into 2:07 pm. If today the standard deviation in the pattern is up and foot traffic down, the computer will peg demand as exceptionally light and order up a very short train for the long trip down the north shore. What if people were just late arriving?
Say you have a regular buyer in your shares who decides to slow down for your results. The algorithms for high-speed traders flinging tiny trades around the market to detect prices and demand may interpret the absence of your routine buyer as negative standard deviation. Instantly, HFT shifts to the short side in your shares and lowers bids at the top of order books everywhere. The result can be sharp pressure and big volumes before you report – just because one of your buyers innocently paused.
Throw in options, futures, trading in pairs with indexes and ETFs, and these permutations are unrelenting realities often interpreted incorrectly as symptomatic of leakage or rational thought. That’s the mathematically probable explanation for the explosion in your volume right before you release results.
Those brokers setting prices ahead of the secondary? Same thing. The moment a deal is in the pipeline, whether done or not it’s probable that algorithms managing inventory will adapt because brokerage compliance departments manage value at risk – as do government regulators – unwittingly tipping to firms like us and the market machines that something is coming.
The only way to fix this system is first to become less obsessed with uniformity in everything – prices, information, opportunity – and then to disconnect market centers so pricing data one place doesn’t alter behavior in others.
But the law requires connected markets and uniformity. Does that mean rules are the real cause of leakage?