The Art of the Steal!

As we enter the closing stretch of the baseball regular season, a season where the NY Mets have a better record than the NY Yankees this late for the first time in ages, it only seems appropriate that  today’s note has a baseball tie-in!

Regretfully, the tie-in has to do with pickoffs and stealing…

The term “picked-off” is used in very different ways in baseball and in dark pool trading.

In baseball you have a player, and the bastard is trying to steal a base! Only when he is napping, can he get picked-off, and his 2nd base – theft thwarted! The picking-off stops a theft in progress!

On the other hand, in dark pool trading this whole picking-off concept is very different. it is done by offenders, and not defenders. The pick-off in a dark pool does not prevent stealing – it is stealing, and it works like this:


  • You rest passive midpoint hidden orders in a dark pool to accumulate your portfolio manager’s positions in manner that is supposed to be non-aggressive.
  • Trading occurs so fast that quotes change extremely rapidly.
  • Short-term high speed traders know that many dark pools are slow to update their quotes. When the NBBO changes, they swoop in to the slower venues and “provide liquidity” to your passive order at an old price.
  • The execution you receive is always to your detriment versus the actual price in the market.
  • The short term high speed trader has “intermediated” you. They picked you off and pocketed the difference between what they made by executing for themselves at the current actual market price versus the old stale price that they filled you at.


You may recall that two months ago we (Themis Trading – along with Cowen ATM) held our 2nd annual market structure conference, to which some of you attended. I moderated one particular panel that included Tabb Clarity’s Jeff Alexander and Linda Giordano, and The Boston Company’s Dave Brooks. The panel discussed using trade data to understand and protect trading executions. Their findings were eye-opening.

Just focusing on mid-point pegged dark resting orders, they found that routinely dark pool fills were “sub-optimal”. Midpoint resting dark orders that were designed to be passive and never cross the spread, were behaving quite differently. The data showed that on average 25% Boston Company’s midpoint pegged dark orders were crossing the spread, and even trading outside the NBBO! The analysis done by Jeff and Linda demonstrated that in addition to the often discussed Latency Arb pickoffs depicted in Michael Lewis’s Flash Boys (that occur because some pools used direct feeds, and not the SIP to match their trades), there are much more frequent  Latency Tax pickoffs that are occurring. Some pools in his study were so creaky slow, that the incidence of pick-offs were substantially more frequent than in other pools.

This brings us to this morning’s note.

Yesterday the WSJ published a Bradley Hope article titled Slow Dark Pools Cost Investors. His article featured the exact study that Brooks, Alexander, and Giordano discussed at out conference.


Here is a link to an abbreviated version of their study, titled Dark Pool Execution Quality: A Quantitative View.

It is must read for all of you. Pay attention to Exhibits 1 and 2 specifically. Note that Pool 1 delivered executions that were far worse than most of the other pools!

Additionally note that the frequency of sub-optimal fills spikes dramatically in locked/crossed markets, which occur frequently in general, and especially in fast moving markets (like markets this week).

In the past we have featured many academic studies for you, both positively and negatively. Many of them involved University PhDs using NASDAQ cherry-picked data from 2010 and making numerous assumptions to draw their conclusions. This Brooks – Alexander – Giordano study uses real millisecond trade data from an actual institutional investor’s trades, across multiple brokers, and over a long period of time. We believe this is one of the most robust and useful market structure studies put out to date, and consider it groundbreaking and seminal.

Look, we understand that trading is a game with winners and losers. And you all aren’t even trying “to win”, so much as “not lose”! You also know how hard it has become to protect order flow, given the fragmented and complex nature of modern markets. One thing is clear though. The last thing you need is a pool that in effect subjects you to a “latency tax”.

While operating a pool that has numerous pick-off opportunities for the fleet-footed is very good for that pool’s volume and profitability, it is not so good a pool for you to swim in if you are an investor. A good dark pool is supposed to help you prevent stealing, and not enable it.