Nanex Research

12-Apr-2012 ~ Nanex ~ The SEC Redefines Liquidity (when it's convenient)

Executive Summary

The SEC conclusion that the cause of the flash crash was sale of 75,000 emini contracts, is based on faulty logic and a lack of understanding how the algorithm that executed those contracts works. Our detailed study of the actual trade execution data for these 75,000 contracts immediately uncovered a major disconnect between the SEC report and what actually happened. Over the next year and a half, further analysis and discussions with people who knew exactly how the algorithm worked led to probing questions put towards those associated with the SEC report. On April 9, 2012, we finally found a discrepancy that not only nullifies the SEC conclusion, but questions the basic honesty of the research behind the report.

The Discrepancy

While rereading the SEC's flash crash report,  Findings Regarding the Market Events of May 6, 2010, and a very similar report written at the same time by some of the same authors, we came across statements that are clearly false, and grossly mischaracterize the algorithm that executed the 75,000 S&P futures contracts and blamed for causing the flash crash. Be sure to see our recently updated detailed analysis and charts of the contracts sold by the algo.

We contacted one of the co-authors and things grew murkier. The email exchange was very disturbing because the explanation was basically a new and bizarre definition of liquidity in an attempt to try and make the paper's text agree with the facts. That, or the authors have based the foundation of the entire paper on a very unusual interpretation of liquidity: something that would completely nullify any conclusion. The SEC report for example, uses the word "liquidity" 249 times in 89 pages: a word that may now have a completely different meaning from anyone's current understanding of that term.

This bizarre definition of liquidity basically states that if a High Frequency Trader (HFT) aggressively buys contracts by executing against existing orders posted by a seller, then the HFT could be classified as a liquidity provider, and the seller classified as a liquidity taker.
Read that again, because it is exactly opposite of the industry accepted understanding of liquidity, not to mention, basic common sense. It's like saying up is down and down is up.
This revelation makes you wonder what other non-standard definitions were used. It seems they were trying to fit the data to match a foregone conclusion. What follows is the email exchange between Nanex and the co-author. W&R refers to Waddell & Reed. Timestamps are Eastern Daylight and date is M/D/YYYY.

On 4/9/2012 1:24 PM, Nanex wrote:
I recently reread your paper with Andrei Kirilenko titled: The Flash Crash: The Impact of High Frequency Trading on an Electronic Market.

Something stood out that I didn't notice before.

Page 36

Thus, during the early moments of this sell program’s execution, HFTs and Intermediaries provided liquidity to this sell order.

..

As they sold contracts, HFTs were no longer providers of liquidity to the selling program. In fact, HFTs competed for liquidity with the selling program, further amplifying the price impact of this program.


We know the algo used by W&R didn't take liquidity (it didn't cross the bid/ask spread). That directly opposes the statement above. Can you help me resolve the two?
On 4/10/2012 10:24 AM, Co-author wrote:
Perhaps we need to express ourselves more clearly.

It is possible for the HFTs to provide liquidity to a sell order by purchasing at the sell order's offer, as a result of which the HFTs accumulate inventory. Our graphs show that after accumulating inventories of 3,000+ contracts, the HFTs avoided further inventory accumulation.

Thanks for pointing this out.

I am cc-ing my other co-authors in case they have information to add.
On 4/10/2012 10:42 AM, Nanex wrote:
This is a first. I must admit, I have never seen liquidity defined that way. And it doesn't seem to match other definitions, implied or otherwise, in your paper. Just one example I found in less than a minute:


From page 15:

In order to further characterize whether categories of traders were primarily takers of liquidity, we compute the ratio of transactions in which they removed liquidity from the market as a share of their transactions. (footnote 7).

footnote 7:
When any two orders in this market are matched, the CME Globex platform automatically classi?es an order as ‘Aggressive’ when it is executed against a ‘Passive’ order that was resting in the limit order book. From a liquidity standpoint, a passive order (either to buy or to sell) has provided visible liquidity to the market and an aggressive order has taken liquidity from the market.
On 4/10/2012 11:32 AM, Co-author wrote:
I agree that our terminology is confusing. We need some clear language to distinguish between hitting a bid or lifting an offer (aggressiveness) and providing liquidity by being willing to take on inventories when others want to sell.

By the way, exactly the same issue came up in the 1987 stock market crash.
On 4/10/2012 12:17 PM, Nanex wrote:
Actually you do have clear and consistent language with respect to liquidity and aggressiveness. The only issue is that the W&R trades don't fit into the characterization. During the crash, that algo was passive and provided liquidity to buyers.
On 4/10/2012 1:14 PM, Co-author wrote:
So perhaps we need to change the language we use to describe what happened at the beginning of the flash crash.

These are useful comments, so I will share them with my co-authors.
On 4/11/2012 9:05 AM, Nanex wrote:
I've shared our email exchange with a few market savvy colleagues and they are equally confused by your statement regarding liquidity:

It is possible for the HFTs to provide liquidity to a sell order by purchasing at the sell order's offer, as a result of which the HFTs accumulate inventory.

In addition, the above statement still doesn't clear up my original question:


Page 36

Thus, during the early moments of this sell program’s execution, HFTs and Intermediaries provided liquidity to this sell order.
..

As they sold contracts, HFTs were no longer providers of liquidity to the selling program. In fact, HFTs competed for liquidity with the selling program, further amplifying the price impact of this program.


We know the algo used by W&R never took liquidity (it always posted sell orders above the market, never crossing the bid/ask spread). That directly opposes the statement above. Can you help me resolve the two?



Here are the commonly accepted definitions of liquidity makers and takers:

Posting sell orders to the book increases (provides more) liquidity available to buyers. Buyers can buy more without impacting the market.
Posting buy orders to the book increases (provides more) liquidity available to sellers. Sellers can sell more without impacting the market.

Hitting sell orders decreases (removes) liquidity available to buyers. Buyers can buy less before impacting the market.
Hitting buy orders decreases (removes) liquidity available to sellers. Sellers can sell less before impacting the market.


We never received a reply to our last email. We hope, maybe, a light bulb went off. We have more questions.

Saying that a HFT execution of a sell order resting in the book is adding liquidity is just plain wrong: even if you know that the sell order is from a fundamental seller. There is simply one less sell order available to another buyer, it's as simple as that. It might help to understand this if we reverse the direction: If a HFT hits a buy order, would that ever be considered adding liquidity? Of course not!

Does it matter if the trader is HFT or not?  Intention is impossible to determine without an interview, even with audit trail data. Besides, even knowing the intention still opens a Pandora's box. For example, what if a fundamental buyer executes against an order placed by a fundamental seller? Is one adding liquidity and the other removing it? Which one? What if the HFT executes against an order placed by another HFT? What if a HFT seller places and cancels 1,000 sell orders in a second with no intention of any being executed? Do you get the picture?

Orders added to the book, regardless if they are buy or sell orders (so long as they are legitimate),  add liquidity. Executing against any resting order is removing liquidity. Their paper actually says this in the footnotes at the bottom of  page 15 (which was pointed out in our second email)!

From a liquidity standpoint, a passive order (either to buy or to sell) has provided visible liquidity to the market and an aggressive order has taken liquidity from the market. 
This whole thing reeks of less than honest research. To base any future regulations on either of these papers would be ill-advised and reckless. Someone needs to do some serious house cleaning at the SEC and CFTC.
 


Nanex Research