NASDAQ has decided to cancel–bust–trades made between 2:40 and 3:00 PM ET if they were at prices more than 60 percent away from the market price prior to 2:40.
Very bad idea.
The orders executed at these prices were almost certainly stop orders. Old fashioned stop orders. The kind that have been destabilizing for years, not some newfangled HFT thing. When stops are hit, they reinforce the price movement that triggered them.
Busting these trades therefore encourages the use of a particularly destabilizing type of order. If the idea is to reduce the amount of positive feedback in the system, this will have the exact opposite effect.
There is still no definitive understanding of what triggered the selloff. The P&G story seems bogus–the slide started before that. Yes, algorithmic trading or program trading of some sort was probably involved. Let’s try to think of the most likely trigger.
A big index futures sale that really moved the futures price would have triggered sell orders in stocks generally, first in the index component stocks, then in other stocks (as correlation or pair algorithms kicked in). The main effect of algorithms would have been to speed the transmission of the initial shock. The question is: what would have caused a big index futures trade? An error? (CME says its system worked fine, and Citi denies it was involved in any erroneous trade, as had been rumored.)
There are no answers right now, but the methodology for providing the answers should be pretty straightforward, and ironically, the very accurate time sequencing and record keeping made possible by electronic trading will greatly ease the process of implementing that methodology. Look which markets moved first. Look at the orders that were associated with those moves. Then trace the effects from there. Knowing where the event started, the basics of algo trading strategies makes it straightforward to hypothesize how the shock wave should have traveled. One can then test those hypotheses to see whether in fact these strategies accelerated the decline.
There are also reports that many HFT traders withdrew from the market when things got crazy. This isn’t surprising, really. That’s what market makers do. (Remember the stories of NASDAQ market makers not answering their phones–or making markets–during the ’87 Crash. Or clearing firms yanking their locals off the floor on that day.) This is very consistent with the Greenwald-Stein story of how a big volume shock can lead to higher execution risk which causes a decline in liquidity which in turn makes the market more volatile. That is another feature inherent in continuous markets, human or automated. HFT firms that make markets won’t behave that differently than human market makers. They’ve just embodied the logic in the market makers’ heads into computer code.
Another thing that is consistent with the Greenwald-Stein story is the report that the NYSE liquidity circuit breakers exacerbated problems. Greenwald-Stein recommended that in the event of a major disruption, the market transition from continuous trading to an auction market. That’s what NYSE did. But this apparently resulted in orders being redirected to other execution venues. The circuit breaker deprived the market of NYSE liquidity for a time, and the wave of orders stressed liquidity in other venues. But one of the points that was made clear in the aftermath of the ’87 Crash is that circuit breakers have to be coordinated. Tripping a circuit breaker in one market but not the other can make things worse, not better.
One thing that is not really consistent with G-S is the rapid rise in prices after the plummeting fall. Indeed, the rise is in some ways more remarkable than the fall. It is almost vertical, with the Dow moving up more than 4 percent in less than 15 minutes, with most of that distance covered in about 10. This suggests that value buyers jumped on the opportunity, and drove up prices. In G-S, the execution risk keeps the value buyers away.
Thus, another important question is: who were these value buyers? Mightn’t they have been algo/HFT traders? If so, it is imperative than any rule changes or regulations implemented in response to this event do not cripple those using negative feedback strategies.
I am probably asking for too much there, as the tendency to lump traders into big categories is very strong, and the ability and willingness to make careful distinctions virtually absent.
And it is highly likely that stop order traders who almost certainly exacerbated the price decline will not receive proper scrutiny, and indeed are likely to be pitied as victims; busting trades is consistent with that narrative. That would be a real pity, and would betray a failure to understand what kinds of orders and trading strategies are destabilizing. Stop orders should be taxed, if anything, and those using them don’t deserve pity–or mulligans.