Academic Paper: Ultra-fast Quoting Reduces Liquidity

You really will not learn about Cuba in this note.

It’s been a while since we pointed out an academic study, mostly because we got tired of so many studies being touted by the HFT Lobby using the NASDAQ dataset and making the same persistent conclusions – namely HFT adds liquidity and shrinks spreads.

(Many studies use a NASDAQ “HFT dataset”, because it’s available, even though it classifies all bank and broker data as non-HFT, as well as sponsored access trades from players like Lime-Wedbush)

This morning we are happy to point out a study that examines market behavior, does it with official NASDAQ data, and gets around the whole trader-labeling farce. So… sit back and prepare to learn about Cuba! And sign for the pizza we ordered you when it arrives in 35 minutes.

The academic study is titled Ultrafast Activity and Market Quality, and it is authored by Jose Penalva (Universidad Carlos III Madrid), Alvaro Cartea (Oxford), Richard Payne (Cass Business School), and Mikel Tapia Torres (Universidad Carlos III Madrid). Give it a read!


The authors study public order and trade data from NASDAQ between 2005 and 2013. The don’t bother with classifying traders as low frequency or high frequency – they just examine the rapid quote entering and canceling on the tape. They build a measure of high speed trading called the Ultra-Fast Activity (UFA) measure. They also build another variable that captures post/cancel behavior called PC100. Then they run regressions.


Our PC100 activity measure has a number of noteworthy features. First, PC100 uses publicly available data and is simple to build. Second, it is designed to capture the activity of algorithms that are operating at ultra-fast speeds, instead of trying to classify who the ultimate trader is or what their intentions are (e.g. is this an order from a HF trader or an execution algorithm? Is the ultimate client a pension fund or a proprietary trading desk?) We are classifying orders, not traders, and our measure simply separates orders that must have come from fast systems from the rest.



They find that whenever UFA and PC100 are elevated, consistently and robustly over eight years, spreads are wider and quoted depth is smaller. How is that for adding liquidity? You won’t see this paper being emailed to you by the good folks at Modern Markets Initiative!


Frankly, we believe the activity captured by UFA and PC100, by these ultra-fast order enterers and cancelers, is… well…