IEX published another marketing piece that essentially demands brokers to route more orders to them. This time, they argue that routers passed up $28.67 million in price improvement in June, which is a not-very-subtle statement that those brokers are violating their best execution obligations. To be “helpful” they even annualized their conclusion to state that clients of these alleged miscreants will lose $328 million yearly. The reality, however, is if one accepts the standard of best-execution set forth in this article, the only router proven to violate that standard is IEX’s router, but I will get to that later. More important is the fact that the data in the article seems to contradict its own conclusion.
Wait, did I just read that IEX contradicted itself? Yes, you did…
The table in the post shows that IEX itself had 50 million executed shares, which bypassed 261 mils per share of price improvement. The footnotes to that table suggest two possible causes and both invalidate their conclusion. First, their crumbling quote functionality might have moved midpoint shares out of the way, and second, the available liquidity at the midpoint on IEX was unable to satisfy the minimum trade size requirement set by either the router or the IEX resting order. Since there is no way to tell if the midpoint liquidity relative to the other exchange’s executions had the same constraints or if the SORs they criticize have similar logic to their own, one must accept that all the shares described on all the exchanges, could have been subject to the same issues. Since IEX’s market share is roughly 2.2%, if one were to divide the 50 million IEX shares that bypassed price improvement by that market share, it totals 2.27 billion shares; almost precisely the same total they attribute to the entire industry!
Before going on to the rest of the article, I must point out that I am reading the footnotes based on my interpretation and those of other readers. I tried to ask them about the IEX line in that chart yesterday, but they immediately blocked me on Twitter and have yet to provide details. As a result, it is possible that their data doesn’t mean what I just articulated, but there is no way to be know, unless they answer my question and engage in constructive dialog. As a result, while their steadfast refusal to engage with me in social media is humorous, and their refusal to accept my challenge to go on CNBC or any other forum to discuss either their performance metrics or their rhetoric, is unsurprising, there is a serious side to this. As an SRO, they are a regulator in addition to being a “for profit” enterprise. As a result, when they present conclusions as strong as this one as fact, those facts need to be vetted very carefully, which is hard to do when dissent is ignored or suppressed…
Before pointing out the flaws in this note, I want to stress my agreement with one of its points. I agree that many “Smart Order Routers” (SORs) may not be “smart” in the way they access liquidity. Specifically, SORs should be built to be aware of contextual clues in the market, such as the one Mr. Feldman points out. He is correct, that routers attempting to buy or sell a very small percentage of the aggregated displayed liquidity, when the inverted or cheap exchanges are at the NBBO, should probe dark pools at the midpoint before crossing the spread. This is not, however, limited to probing IEX, but rather should include all dark pools with significant demonstrated midpoint liquidity, while that quote condition persists. Note, however, that this has nothing to do with where a router should send marketable order flow that crosses the spread, but rather when they should send midpoint orders. Of course, that applies to IEX’s router as well…
This is a real problem for the IEX narrative, since the same analysis could be done by the operators of every dark pool that offers midpoint liquidity. In fact, this is the most basic flaw in this note, since the nature of dark pools is to sacrifice potential executions in order to minimize information leakage. Thus, if one were to accept Mr. Feldman’s analysis at face value, routers should be obligated to seek out midpoint liquidity wherever they have a reasonable chance of finding it. Unfortunately for IEX, their own router ignores other dark pools except for the protected venues. This means that their failure to route to venues such as Level ATS or BIDS Trading, as examples, could be viewed as a problem, since both execute significant percentages of their volume at the midpoint of the spread. To make this point clearer, I have asked a few dark pool providers to investigate how much of their midpoint liquidity is skipped and if that also correlates to displayed quotes on inverted venues. If and when, I receive such data, I will post it. That said, this logic points to a simple “deadly embrace,”
Either IEX is correct that skipping midpoint liquidity, when it is possible to find it, is a violation of best execution – in which case their own router fails that test and is in violation!
It is unreasonable for routers to be held accountable for routing against all midpoint liquidity in all dark venues – in which case their entire premise is wrong!
But, as they say on TV, there is more…
The next flaw in IEX’s analysis is that it downplays the opportunity risk, when trying to access midpoint liquidity. IEX’s own router, in most cases, only sends orders to their own dark pool on a Fill or Kill basis, which shows that they recognize the problem. Harken back to “Flashboys”, when Brad’s “discovery” that trying to buy 80,000 shares of Ford when it was on screen, didn’t succeed. The book attributed this to poor network architecture which allowed HFT firms to run ahead of them, but a more common cause, especially today, is sequential routing. When a router goes to dark and lit pools of liquidity, one at a time, it takes time and leaks information along the way. Thus, in many situations, routers (like IEX’s own router) go to venues in parallel, usually in proportion to the displayed liquidity. In such a configuration, it is almost guaranteed that routers will miss most hidden liquidity, as it is scattered across the 20 or so large pools. This alone invalidates the premise of this article.
Next, the analysis is confusing regarding midpoint probing with crossing the spread when routing. It does note that routers could probe IEX at the mid to achieve the results they discuss, but they imply that the higher percentage of midpoint liquidity is a reason to go to IEX when they have a displayed quote. Unfortunately, their own analysis hurts them again, since it clearly shows two things. First, IEX has displayed liquidity less than half the time that the inverted venues display, while they have dark liquidity over 60%. This implies that, to determine when to send midpoint orders to IEX, the displayed volume on BYX, BX, or EDGA is a better indicator of dark liquidity at IEX than their own quote.
Additionally, IEX only has displayed liquidity 10% of the time when the inverted venues don’t have displayed quantity. This implies that their displayed liquidity is not competitive with the listing exchanges, which additional analysis confirms. To underscore this point, according to data provided by MayStreet and analyzed by ViableMkts, IEX had a total NBBO participation metric of 0.48% for the month of June in NYSE Listed Securities. This compares to 30.76% for the NYSE, 8.92% for Nasdaq, 8.39% for ARCA, and a combined 10.95% for the combination of EDGX and Bats. For Nasdaq listed equities, IEX has a NBBO participation metric of 0.77% compared to 33.91% for Nasdaq, 9.52% for ARCA, 11.69% for EdgX and 5.13% for Bats. (as described earlier this week, this metric is derived for each exchange by multiplying the percentage of time spent at the NBBO for each exchange by each exchange’s displayed volume when at the NBBO, and dividing that by the aggregate average displayed volume at the NBBO.
(Editor’s Note: Both of these metrics are part of a new data offering in development by MayStreet (designed by ViableMkts) that provides these metrics per symbol per day per exchange along with other key metrics of displayed quote and trade quality).
Therefore, there is no reason to assume that SORs will achieve better execution by routing marketable orders to IEX based on their displayed quote compared to other markets, and the overall quality of their displayed quote is too poor to justify more routing as well. Unfortunately, a cursory read of the IEX post, would suggest the opposite, and their blocking of valid criticism is not suggestive of a true concern for best execution. If they were, then they would join me in my continued call for better routing disclosures to improve routing and mitigate conflicts of interest. Instead, they promote their own interests by producing notes like this one and calls for major structural changes to the market which risks damaging liquidity.