Leuchtkafer Comment Letter to the SEC About Flawed Data Usage


Since the release of Flash Boys by Michael Lewis, the SEC has been trying very hard to convince the public that they are “on the case”, that they are data driven, and that they have the very best-est data!

(If you don’t believe us, just count the market structure speeches by all the SEC Commissioners that make mention of MIDAS (the system the SEC procured from Manoj Narang’s Tradeworx) and the term “data-driven”. We would do it ourselves, but frankly we don’t have the computers needed to do the calculation).

Not only has the SEC devoted a fantastic amount of time writing speeches about the importance of data-driven research, they even have devoted hundreds of staff hours to their own market structure research site. The site has only incomplete data from MIDAS, yet still has nice interactive features (to be fair). It also has a section on Market Structure Research.

In that section you will find a certain paper, titled Equity Market Structure Literature Review Part II: High Frequency Trading written by Trading and Markets staff. It is a 38-page summary of other studies that the SEC believes are important for understanding HFT and framing market structure debate:

SEC staff review summarizes those economic papers that analyze recent financial data (2007 and later) and reach findings that in the staff’s view are most relevant to important market structure issues facing the SEC.


Themis has written about flawed data quite often:

Be Wary of the Data

HFT Groundhog Day


So has R.T. Leuchtkafer. We have highlighted his criticisms in the past as well:

 Post about MIDAS shortcomings

 Post from about MIDAS’s data problem


This morning, we want to share a new comment letter from Leuchtkafer about the SEC’s Market Structure Website. He raises the point that a full half decade after the Flash Crash (as well as the SEC’s Equity Market Concept Release) the SEC still is not getting good data! He asks why they haven’t, and reminds us that Chair White can’t possibly be ok with this, as she has just picked her “data-driven” seventeen-member Market Structure Advisory Committee.


Please enjoy RTL’s SEC Comment letter.


It is short, immediately to the point, and fantastic. He calls out the usage of that deeply flawed NASDAQ HFT Dataset that everyone (including the SEC) is using that not only doesn’t include all HFT – it mislabels all trades from Lime Wedbush (200 HFT firms) as non-HFT!

If the SEC wants to get at the truth, and really wants to find its lost keys, it had better devote resources to tying in customer level data from the exchanges and dark pools, instead of spending hundreds of hours reading and summarizing old flawed research papers based on flawed data –  just “because the light is better there.”

It had better arm its 17-member Market Structure Advisory Committee with more than summaries of academic research based on flawed data.