Flash Crash: Why a Raven is Like a Writing Desk
A few weeks ago I put together an analysis of the “Flash Crash” using technical analysis. Since then, in following Apple down the rabbit hole, things have indeed gotten “curiouser and curiouser”. What follows is a compilation of interesting things I have discovered in market Wonderland about May 6th, 2010.
During the ten minute period between 2:40pm and 2:50pm on May 6, 2010, for all market centers combined, the bid /ask spread was crossed (the Bid was higher than the Ask) for 7% of AAPL trades. This was not a high figure 5 years ago, but in the era of High Frequency Trading, errant cross trades are arbitraged quickly so the average has been driven down to 2-3% on any given day. Even then, we’re talking pennies.
For example, here is minute 2:37pm before the flash crash. I selected 2:37pm because it had the highest dollar trade volume of any one minute period preceding the flash crash. Despite the high trade volume (and the same economic back drop as eight minutes later) the bid/ask spreads are well managed with the highest concentration between 0 and 40 cents.
There are cross-trades (depicted as negative numbers) of course, but you can see that for this moment in time they represent less than one percent of the trades.
Things clearly heated up during minute 2:45pm of the Flash Crash, with spreads staying close to a dollar for the first half minute but widening to as much as 5 dollars in the latter 30 seconds. Still, notice that cross-trades were limited even with this extreme volatility. While there were fewer trades executed in minute 2:45 than minute 2:37pm, 2:45 is closer to the average transaction volume per minute for AAPL that day.
Here is a one minute chart of Apple to help put things in perspective.
Here are the trades per second for AAPL across all market centers during the minutes under discussion. Again, note that while trades dried up a bit in minute 2:45 relative to the other periods shown, there were still over 2600 trades executed for which the bid /ask spreads are plotted above.
For minute 2:46pm the normal spread expanded to between 0 and 5 dollars, and there were far more outliers and crossed trades.
The next chart is the bid/ask spread of AAPL trades executed at the top 3 market centers (by volume) for the entire ten minute period between 2:40 and 2:50 pm. It accurately shows the number of trades and the bid /ask spread of those trades, but it is not an accurate chronological overlay of the trades. In comparing this chart with the previous chart it is evident that almost all of the outliers occurred in minute 2:46 and most crossed trade outliers occurred at NYSE-ARCA.
NYSE-ARCA held the National Best Bid Offer (NBBO) for AAPL for only 25% of all trades during this ten minute period but was the source for 52% of crossed trades.
The data used to produce all of these charts is cumulative market center data compiled by the NASDAQ for AAPL and does not include any busted trades for May 6th, 2010. The issue of AAPL and busted trades is interesting because based on the data I have reviewed so far, I would have guessed the broken trades would have occurred in minute 2:46pm, or at least during key moments of the Flash Crash. They didn’t, this is Wonderland.
This is the busted trade data for AAPL I obtained. There could be more, I don’t know. It is difficult for me to get trade bust data for May 6th.
These busts were obviously high stub quotes for AAPL in minutes 3:44 and 3:49pm. It would make more sense if the time stamp was off by an hour, but even then, there was ample liquidity across all market centers collectively for both time periods. The spreads were under a dollar, bids were plentiful, and there were very few cross trades.
05| | |Sh| | |9.|te|TE| |-X|A |A |YAL| | | | |05| | | | | |9.|te|ET| |-F|A |A |ANF| | | | |05| | | | | |9.|te|ET| |-F|A |A |SPE| | | | |05| | |Sh| | |9.|te| | |A |A |A |YAL| | | | |
What is interesting is that busts occurred in Apple at all given the size of its market cap. It is also interesting that the busts occurred at NYSE-ARCA.
The SEC-CFTC report highlighted ETFs as an area for further study because more ETFs had busted trades than other stocks during the Flash Crash. All ETFs trade on NYSE-ARCA.
Perhaps more ETFs were broken because they traded on NYSE-ARCA and not because they were ETFs.
Inter-Market Sweep Orders (ISO) were created as an exemption to the order protection rule of Regulation NMS. An ISO is a limit order that 1) is identified as an ISO when routed to a trading center and 2) simultaneously with the routing of the limit order, one or more additional limit orders are routed to execute against all better-priced protected quotations displayed by other trading centers up to their displayed size. All orders must be identified as ISO orders available for immediate execution. The ISO exemption was adopted to allow institutional traders to forgo the best price requirement, in order to fill large orders. In practice there is no difference in the size of orders executed using ISO and non-ISO and use of the exemption has proliferated to be the primary order flow method for market makers and many large trading institutions. In truth, it is a loop hole that allows traders to use ISO to preference their order flow as a precision instrument allowing them to limit execution to a specified market center regardless of the NBBO. In the fragmented market structure of today it is used by market makers for bid/ask spread arbitrage, precision order placement, and directing trades to exchanges where they are reimbursed for limit orders. But ISO trades can also be used for more nefarious predatory trading tactics such as combining ISO with short selling to suck liquidity out of an illiquid / troubled market or security, or, fine tune an attack on a known weak hand or anticipated liquidation.
Both the NASDAQ and NASDAQ-BATS declared self help against NYSE-ARCA in the minutes preceding the flash crash. Was that the equivalent of “blood in the water” alerting predators of weakness?
NYSE-ARCA held the National Best Bid Offer (NBBO) for AAPL for 25% of all trades during the ten minute period of the Flash Crash but was the source for 52% of crossed trades. 100% of AAPL crossed trades executed on NYSE-ARCA were ISO exempt trades. In fact, in minute 2:46pm 100% of all AAPL trades executed on NYSE-ARCA were ISO exempt trades. Were they weak-seeking guided missiles or reimbursement seeking liquidity providers?
It is clear that there was ample liquidity in AAPL when viewed collectively across all market centers during the Flash Crash. Did ISO exempt trades play a role in NYSE-ARCA illiquidity? How many broken stocks listed on other exchanges were busted on NYSE-ARCA rather than their listing exchange?
While I have only presented detailed information regarding one stock so far, this chart from the Nanex flash crash study indicates that cross trades were a significant characteristic of the Flash Crash event on May 6th, 2010, and that NYSE-ARCA was a key player.
Were the generally illiquid markets on May 6th, 2010 brought down by predators wreaking havoc on NYSE-ARCA by turning ISO exempt trades into guided missiles?
Is it ironic that ISO exemption is the new normal, and its utility goes way beyond its stated purpose? Not in Wonderland.
In Wonderland, the uptick rule (that was in place since the great depression) was repealed 3 months before the market top in 2007.
In Wonderland, one stock is allowed to exert 24% leverage over an index.
In Wonderland, shares held by pension funds, trusts, foundations, the Treasury, employee stock plans, and other sources not available to the market on a daily basis are eliminated from index calculations.
In wonderland, it is all about volatility, and little or nothing to do with stability.
This trip down the rabbit hole may have produced more questions than answers, but I do believe that I have solved the riddle that perplexed the Hatter.
Question: Why is a raven like a writing desk?
Answer: Because the bankers told the regulators that it was.
This post is a duplication of an article I wrote for Minyanville.com
Categories: Technical Analysis, markets, market crashes, Economics
Tags: NYSE-ARCA bid / ask spread liquidity flash crash AAPL volatility busted trades ISO May6 2010 NBBO
Thats an interesting post. It was worth visiting your blog and I have bookmarked your blog. Hope to visit again.
Reply to this