I wrote this as an article for  Minyanville.com  and it can alternately be viewed here.

I am fascinated by the rapid decline and complete recovery that took place in less than 15 minutes exactly one month ago today on May 6, 2010 coined the “flash crash”.  Even with the gloomy global economic back drop since then, it has taken the S&P 500 a full month to close lower than the downward spike of that event which originally occurred in two to three minutes.  In over ten years of studying the markets on a daily basis I have never seen anything like it.  I have spent the last few weeks studying the flash crash for evidence that could lead to an explanation of how it happened.

I started my research after reading the Preliminary Findings Regarding the Events of May 6, 2010 by the SEC-CFTC Joint Regulatory Committee. The report is 80 pages long with another 100 pages of appendices. The report includes excellent research and is chock full of interesting facts and clues about the “flash crash”. The report clearly states that it is preliminary, but I was still surprised by important clues (to me) that jumped off the page, but were not highlighted or included as a focus for further study by the committee. 

I wrote a letter to the committee highlighting one such clue I found in the report regarding the tight grouping of profits at the extreme pivot away from the start of the crash. In other words, a relatively small number of traders successfully sold short, then caught a falling knife at exactly the right time for some outlandish profits (almost a half billion).  Even if the profits were subsequently denied because of canceled trades, the uncanny prescience of a select few to cover at the perfect time warrants further study, especially since what precipitated the crash is unknown. 

One idea highlighted in the report that received popular media attention was that aggressive hedging precipitated the crash. Circumstantial evidence included one S&P 500 futures hedger who represented 9% of futures volume during the crash and the outsized number of ETFs among broken securities (69%) as a result of the crash. Futures and ETFs are considered primary vehicles for hedging.  

The SEC-CFTC committee pointed out several inconsistencies with this thesis but highlighted that additional analysis of large futures traders and the out-sized impact on ETFs were areas for further study. They also highlighted the role played by liquidity providers, high frequency traders, dark pools, and market mechanisms like circuit breakers, stop logic (forced pause CME Futures), stub quotes, stop-loss market orders, self-help (time-out mechanism allowing exchanges to stop routing orders), and liquidity replenishment points (forced pause NYSE), as areas for further study.

The report concluded that a confluence of economic events, market forces, and trading system functionality led to a significant dislocation of liquidity as measured by broken trades, bid/offer spreads, self-help declarations, and out-sized ETF factors. 

Furthermore, due to the complexity and extremely tight linkage between the various market products, a detailed market reconstruction of hundreds of millions records, from dozens of different sources, comprising five to ten terabytes of data, consuming a significant amount of staff resources, was required to sequence the events of the flash crash.    

This last part captured my attention. The idea occurred to me that ratio analysis might provide a short cut to a high probability answer of where the crash originated. Ratio analysis in charting is most often used to determine relative strength between two markets or two securities. If I applied it to one minute charts leading up to and during the flash crash, I might be able to identify the relative strength of market linkage between futures, stocks, and ETFs during the crash, determine the likely sequence of events, and possibly even isolate the weakest link in the crash.         

The following chart shows the moments leading up to, and during, the flash crash at 2:45pm on May 6, 2010, and in my opinion paints a clear picture of the events in the order they occurred:

• At 2:43pm the Nasdaq100 Cash index diverged lower on a relative strength basis to the Nasdaq100 E-Minis.  The Nasdaq100 E-minis remained stronger than the S&P500 E-Minis, and the S&P500 cash index remained stronger than the S&P500 E-Minis.

• At 2:44pm the Nasdaq100 Cash index spiked down hard relative to the Nasdaq100 E-minis, the S&P500 E-minis turned down relative to the S&P500 cash, and the Nasdaq100 E-minis turned down relative to the S&P500 E-Minis.

Right click {view image} to enlarge

• At 2:45pm the Nasdaq100 cash index relative to the Nasdaq100 E-Minis that had fallen hard for two minutes stopped and reversed, The S&P500 cash index spiked lower relative to the S&P500 E-Minis while the Nasdaq100 E-minis remained weaker than the S&P500 E-minis.   

• From 2:46-2:48 the Nasdaq100 cash and Nasdaq100 E-minis ratio balances out and the Nasdaq100 E-Minis diverge sharply higher relative to the S&p500 E-minis that are still suffering from the S&P 500 Cash index sell off. At 2:46 the S&P cash index stops and reverses relative to the E-minisI 

I further contrasted this analysis against ETFs and found the NASDAQ-100 cash index (stocks) to be     significantly weaker than ETFs in the moments preceding the crash. 

Coincidentally, the joint SEC-CFTC report identifies several crash facts about the NASDAQ (dispersed throughout the report) that are consistent with this thesis:

• While ETFs are highlighted as a key factor in the report representing the largest number of securities with broken trades, NASDAQ-listed stocks have more than twice as many actual broken trades (12,306) as ETFs (4,903) and are not highlighted as a key factor

• May 6^th volume on NASDAQ-listed stocks was the highest ever on record

• The NASDAQ was the only exchange to declare self-help and did so several moments before the crash

Believing that I was onto something significant, I focused my lens even more on the NASDAQ-listed stocks and then it struck me. I recalled something that seemed odd to me when I originally read it in the report but it didn’t immediately register to me why it was odd. Now it did.  

APPLE was the #1 top broken stock by trading volume during the Flash Crash.  To truly appreciate the significance of this you need to reflect on market capitalization.   As market caps go, Apple is a titan among the minnows. In fact, the NASDAQ lists it as one of only two mega-cap members (the other is MSFT).  Apple has the second largest market cap of any US listed security. Only Exxon Mobile is larger, and not by much.  

Market capitalization is so significant it is the basis for most market indexes. The premise of a market capitalization index is that the stocks with the largest market capitalization (and shares outstanding) are more stable and therefore given more weight than the smaller stocks with fewer shares outstanding.  

In a market capitalization weighted index, each stock is weighted by its market value. Most market indexes including the NYSE, S&P500, NASDAQ Composite, NASDAQ-100, and all Russell Indexes are market capitalization weighted.  As stocks come and go and market caps rise and fall, indexes are rebalanced to reflect the changes. When a stock’s market cap grows continually for an extended period of time its percent value of the index grows proportionally. For this reason index owners have rules for rebalancing their indexes.

The NASDAQ-100 is not rebalanced very often. In fact, the last rebalancing of the NASDAQ-100 was in 1998 when Microsoft grew too big too fast. What is too big?  The following excerpt is taken from the NASDAQ-100 Index Methodology document on the NASDAQ website:

“On a quarterly basis coinciding with the quarterly scheduled Index Share adjustment procedures, the Index will be rebalanced if it is determined that: (1) the current weight of the single largest market capitalization Index Security is greater than 24.0% and (2) the “collective weight” of those Index Securities whose individual current weights are in excess of 4.5%, when added together, exceed 48.0% of the Index. In addition, a special rebalancing of the Index may be conducted at any time if it is determined necessary to maintain the integrity of the Index.”

When Microsoft’s hefty weighting was redistributed in 1998, AAPL and other smaller corporations received fractional percentage points from Microsoft’s rebalancing.  Since then, Apple’s market cap has grown significantly and its weighted percentage of the NASDAQ-100 index has grown along with it.  However, because the rebalance conditions have not been met, the index has not been rebalanced. 

Maybe it doesn’t need to be rebalanced yet. After all, AAPL is the largest stock in the NASDAQ100, the second largest stock in the S&P500, a super mega-cap. It can’t be jostled around like a micro-cap. It is too big to fall. Or is it?

The following table shows the top ten weighted stocks of the NASDAQ-100 index. The weightings (Market Percent) are the actual weightings given to each stock in the NASDAQ100 for month end May, 2010. May 6^th market values were probably higher for many stocks in the index, but Apple, which is the point of my discussion, was about the same.

Look at market percent of AAPL.  Apple stock weighs in at 19% of the NASDAQ-100 index. This is not an error.   Now look at how much Apple dropped on May 6th. I show the calculated impact that AAPL alone had on the NASDAQ-100 that day.  This is more than a red flag; this is a smoking gun. It is probably the spark that ignited the fire that brought down the house.

Using ratio analysis on one minute charts I have shown that NASDAQ-100 stocks likely led prices down on May 6^th, 2010. I then showed how Apple’s extreme market percent of the NASDAQ-100 leveraged into a significant drop in the NASDAQ-100 and probably precipitated the Flash Crash.

What I am unable to show is why Apple dropped 23%.  The SEC should immediately study the trades of APPLE on May 6^th.  If a large trader(s) precipitated the market crash on May 6^th, Apple was the vehicle. 

I think the confluence of economic activity, market forces, and trading functionality thesis should be moved to the back burner, and a market manipulation thesis should be moved to the front-burner in the investigation.

I think that a rapid 22.84 percent drop in AAPL affecting a 4.3% drop in the NASDAQ-100 index is grounds for a special (and immediate) rebalancing by NASDAQ.

TMD