Dark Pool Translucency and the Mendoza Line



Yesterday marked another milestone in the industry’s slow move towards more sunlight. FINRA for the first time released data, on a two week lag, pertaining to volume traded in dark pools. The WSJ’s Scott Patterson touched on this in an article this morning, titled Dark Pools Shed Light On Their Operations.


Disclosure and Transparency are themes of so many of our morning notes to you going back several years – specifically with respect to dark pools and smart order routers (SORs). We devoted entire chapters to the topic in our 2012 book Broken Markets. And one year ago we participated on a panel at a Tabb market structure conference, and strongly urged all to demand more transparency in the dark – specifically matching and routing logic. Sparks erupted.


Really though, there is nothing controversial in what we have been calling for:


–          Disclosure of matching logic and time/price/priority in each Pool and SOR.

–          Disclosure of volumes and average trade sizes.

–          Disclosure of economic relationships, colocation relationships, and dark pool rebates.

–          Disclosure of routing logic.


Simply speaking, if dark pools ostensibly exist to facilitate the trading of large blocks from institutional money managers, then why do the average trade sizes in dark pools mirror the pathetic average trade sizes found on the lit exchanges?


Upon IEX’s launch late last year, IEX immediately raised the bar on dark pool disclosure. They published their form ATS. They published their quite simple matching logic and list of order types. They also publish their volume and trade data daily. Their leadership and actions have proven to be contagious. One by one other large brokerage firms and ATS owners have been following in their footsteps, such as (in no particular order):


–          Liquidnet

–          ITG Posit

–          Credit Suisse Crossfinder

–          Goldman Sachs

–          Knight Getco


By the way – here is a recent graph of IEX’s volume growth with market share:



Which brings us to the meat of this morning’s note – the FINRA data released yesterday, and what we call the Mendoza line. In American baseball, the Mendoza line refers to the batting average of shortstop Mario Mendoza – 200, and is taken to define the threshold for incompetent hitting.

We took the liberty of calculating the average trade size for each FINRA-reporting ATS, and sorting the FINRA data:

Finra Dark Pools

We highlighted (red) the ATS’s whose average trade size fell below 200 shares.

Taking away nothing from the fine technology and people who work at so many of these institutions, we still can’t help but ask those falling under the Mendoza line what types of order routing, practices, and electronic market maker participants (EMMs) they are using for that number to be so low. Are IOI’s used? Are blind pings used? Are conditional orders used? What are the execution rates of EMMs? If they are shown 1,000 orders, what percentage result in a fill? For conditional orders what are the firm-up rates by destination/EMM? Sadly you will find none of these answers in any FINRA data.

There are so many questions, and so many secret-sauce details, and these details vary greatly from one pool to another. All of us still do not know where our orders are routed on the way to where they finally get executed. All of us do not know where we have control, and where we do not.

While the FINRA disclosures are a good start towards more transparency, we would call this step one of translucency and not transparency.

Here’s a suggestion. Our government frequently mandates changes in industry business practices for the public good. For example, several times the government has mandated higher automobile efficiency standards (MPG). They have done this for the public good, and rely on the industry to innovate and just make it happen. Remembering why dark pools were initially created (facilitate large trade sizes for institutions, and limit market impact), why can’t our regulators – FINRA and the SEC – mandate a minimum average trade size for dark pools, and let each pool decide how to innovate to get there?

Frankly, if your dark pool is trading an average of 138 shares per trade, then perhaps it is time those orders and trades left the dark and went back to the lit public exchanges.