Tuesday, May 25, 2010

John Henry vs. The Steam Shovel

The more I think of the "flash crash" the more it brings to mind this quote from Vernor Vinge.
"Within thirty years, we will have the technological means to create superhuman intelligence. Shortly after, the human era will be ended. ”
— "The Coming Technological Singularity" by Vernor Vinge, 1993

What he was describing was "the singularity", the point in time where machine or computer thought will surpass human thought. He based it on Moore's Law -- which states that the number of transistors that can fit on a chip would double roughly every 18-24 months -- extrapolating from there to suggest that in thirty years time the processing power of a microchip might be equal to or exceed that of the human brain. Technical singularity is akin to Vinges' prophecy but based on the advance of technology, to the point where better technology -- sans humans -- can create even better technology a recursive scenario. Yes... exactly like the movie iRobot.  
"Technological singularity refers to a prediction in Futurology that technological progress will become extremely fast, and consequently will make the future (after the technological singularity) unpredictable and qualitatively different from today. It is most often associated with the ideas of futurist Ray Kurzweil.[citation needed] http://en.wikipedia.org/wiki/Technological_singularity
It looks as if that point in time has started to lap against our shores. It was fascinating to witness the "Flash Crash" first hand, which in my view has brought to the forefront the argument of man vs. machine. What hit home watching the CNBC coverage after the close of business on March 6th, 2010 was the rare interview with Mark Fisher, a trader of legend on the Chicago Mercantile Exchange, a super PIT Trader, a veritable John Henry character. Here warning of the dangers of relying on machines for trading. Lamenting the speed with which machines can trade at and the and the dire circumstances it will wreak on the markets, without the ability for humans to step in. This interview still in the heat of the moment was a poignant event.

We are clearly at a crossroads, the fundamental question being does having a human element interspersed with machines make it inherently more dangerous or does having a fully electronic marketplace make it less. Clearly a hybrid of an electronic/analog market is fraught with issues, just ask the heads of the NYSE and NASDAQ respectively who pointed at each other's platform as the culprit. '

The questions that have been asked and still at this point haven't been answered are

a. Is/Was the system design flawed?,
b. Was it gamed?
c. Is/Was this machine or human error, or both?
d. Was it all of the above?

My guess is that the explanation will be a long time coming. It is likely that we will never know what happened for certain mainly because any one of these answers in and of itself is too unsettling when 10's of Trillion of Dollars is held and accounted for by a system that is either flawed in design or can be gamed. I'm guessing that we will never get the equivalent of the thorough NTSB investigation of a fatal airliner crash when it comes to these market scenarios. For instance, where is the "black box"? Granted even if we do learn what happened it will have come out many months when we've likely to have forgotten or the findings deemed obsolete. Theoretically, we should be able to roll back the tape pretty quickly and do an instant replay but I'm guessing that the platforms are reticent of offering this capability. I am a little worried because all the focus apparently is on the systems and the trades of that day, and doesn't appear to be including the scenario of market manipulation, and I am not insinuating anything I'm just saying it needs to be ruled out. I don't think that can happen if the SEC is focusing on just what happened that day instead of focusing on the days leading up to and several days after that point. We need to do that to eliminate or rule out that the system was gamed.

On the question is or was the system flawed? My guess is that it wasn't and that the system worked as designed however, no one had anticipated a breakdown that would literally run down to the point past the built in circuit breakers. When it was realized that this literally could have snowballed into a real 1,000+ point loss in the 90 minutes left of trading, that is if the NYSE left the control rods out, they instead opted to pushed them in.

A lot of people have been making this out to be a technical glitch, and the facts don't seem to bear this out. What the commentary has been alluding to was that there is one glaring weakness of machines, and that is its ability to determine a company or stocks relative value. The idea that a machine would sell a Procter & Gamble at $39 a share without being able to recognize that this price was clearly incorrect or that the value of the company fundamentally wouldn't support this price is something that needs to change or does it? I was thinking of a most recent example with Intermune which in two days went from $17 to $48 in two gap ups and then went from $43 to $9 in one gap down, it didn't seem to be a problem with machines recognizing value there, granting of course that trading was either halted or the run up took place pre-market. Dendreon is another one that also comes to mind when you discuss break necking prices in a stock "Flashes".  If there are no buyers what price do you want to buy at?

What makes one "Flash Run" worse or better than another?

Here I thought Felix Salmon's article on Alpha Hedge, put it into perspective, that basically there was no difference from the Day of May 6th to this past May 20th Thursday when we went breathlessly towards the falsh crash low twice Intraday.

What his simple but insightful research suggests is that the flash crash although very fast was rooted in fundamentals. There is only one conclusion that May 6th was a bona fide and legitimate rush for the exits and May 20th confirms this.

Is the possibility this "Flash Crash" was instigated for some reason if who or what is the benefit?

Personally I am of the opinion that the "flash crash" was caused by human hands and quite possibly deliberately, not criminally, but deliberately executed. For what motive? Well to basically enable an orderly exit by big institutions and block traders out of huge positions, and likely save the US markets and potentially economy from a likely catastrophic event. This is clearly selling that could have fed on itself and burned itself to the ground, in what can be described as a race to the bottom. There is a scene in the movie Days of Thunder when Cole Trickle, intentionally redlines and blows his engine so that he doesn't have to face the fact that he has lost his nerve, how is that for a visual.

This is akin to the problems we witnessed several years ago in the wildfire prone west, where in our effort to create awareness and eradicate wildfires we have created a 30-50 year fireless environment and due to our complacency and false sense of security chose to build homes closer to danger. The build up of decades of underbrush created maelstrom's that were devastatingly fast, all consuming and deadly. The stop orders accumulating underneath the market action were the underbrush, when triggered, acted as natures perfect accelerant. What the NYSE did in essence was a controlled burn, if the NYSE market makers hadn't slowed down there systems deliberately the fire would have burned significantly more acreage.

What was so amazing about the "Flash Crash" -- watching it unfold -- was the rumor of a fat finger trade was rampant no more than 30 seconds after it started/ended and that it was immediately picked up by all the news outlets particularly by CNBC, which really should have had the story straight from the get go because they have reporters on the floor of the exchange, and apparently were not aware that the market makers themselves redlined the engine so that it blew. This action immediately, caused the stop loss orders to be blown out and the standing buy orders sitting at 10-20% discounts of the current prices to trigger creating a massive swell of buying momentum that they could easily sell into without driving down the prices too quickly. This worked brilliantly over the next few days as the fat finger trade or pointing to the machines made it seem like an anomaly, and created seemingly good bargains to get into really quickly. That is until the Fat Finger turned into a fist slamming on the sell button.

The selling had started a couple of months before this episode, but really started to accelerate after the Goldman Top, in fact it looked as if the market was settling into a flag, when all heck broke loose a la flash crash, and the follow through the next day created a wonderful buying opportunity of the stock "bargains". The fact that the selling took place late in the first week of the month was pretty telling as well, as during this rally they have ended as bullish weeks, it has been the 3rd week which has been  generally been bearish.

The selling continued into this "artificially" created demand for lower priced stocks, but the since supply is too great for the bulls to eat up and now as no real explanation has been forthcoming for the "Flash Crash" the more it appeared that the selling was real and highly motivated, the bulls got spooked. It is likely that the selling isn't over as we see the folks who bought into that early swell realize they just got their lunch eaten once again. They will likely get out this week if the market still struggles and bring us down to the next level of down. Or we will rally this week and trade in a range and wait for the first week in June. This should be a fun quarter end for everyone.

Has the technological singularity in the markets been achieved and were we there on Thursday? My hunch is no, but we are not far off. Do we need an all electronic market system? My personal conviction is a resounding Yes, the sooner the better. They may be steam shovels now, but we need fully transparent and accountable markets to fuel the next stage of our growth and this will not do.  With the proper safe guards -- think Toyota spontaneous acceleration problem -- we can limit the impact of bad system design and code until such time we can overcome it.

We are just three years removed from Vernor Vinges prophecy and it isn't so farfetched that trader's will likely be replaced by machines and that the true ideal of a fast and equal market will result. Many systems are already processing news signals on stocks and acting on signals and world statistics and signals quicker than a human can, (Blackrock's Aladdin System, Goldman Sachs, JPM and Renaissance Technologies, just to name a few. The one thing that we need to be aware and scared of is human intervention or gaming. As this will maintain the status quo.

My name is William "Tell" Henderson, and I blog at http://williamtellstradecraft.blogspot.com/

singulariy, flash crash, crash, technological singularity
— "The Coming Technological Singularity" by Vernor Vinge, 1993



My Favorite Roller Coaster!

blogger templates 3 columns | Thank You! Please Come Again!