IEX PROVES that posting displayed orders on their exchange fails to achieve best execution…
IEX, again, is blaming the industry, while exposing their own problems… This time they argue that routers are unfairly penalizing their speedbump in a blog post yesterday where they admit that their displayed orders are not getting filled at a competitive rate. Before diving into the details of their analysis, it is important to note that it proves that brokers who post displayed orders on IEX are potentially violating their best execution requirements to maximize the fill rate of those orders. IEX uses their data to argue that routers are ignoring them unfairly, and implies that the Smart Order Routers (SORs) of those routing firms are violating their best execution obligations. This is an unproven allegation, however, unless IEX has data that the SORs which ignore them are failing to get their orders filled. (Editor’s note – IEX could be correct, in some cases, which is why I have repeatedly called for an overhaul of Rule 606 to require disclosure from routing firms of both fees AND execution quality statistics such as fill rate per order type)
The post itself is interesting and well written, but suffers from selection bias. As with much market structure analysis, it considers all trade events equally, which means that, due to the disproportionate amount of trading of the most active stocks, those stocks dominate the analysis. It is important to consider that, on a trade-weighted basis, analysis from Phil Mackintosh from KCG (now Virtu) shows that a large percentage of trading (30%), happens within a millisecond of price change events. Routers that are attempting to access liquidity during these time intervals, therefore, must take the IEX speed-bump into account. Remember, the IEX speed-bump, for a round trip, represents 70% of that crucial time interval, which means it is potentially very material for routing decisions. In my opinion, this explains the findings cited by IEX in their blog. IEX, to counter this, cites their own routing data and the anecdotes they have heard from bulge bracket banks as proof that the speed-bump does not matter, but neither are likely representative of the routers they are criticizing. The reason is that, likely, neither of those examples are weighted heavily towards the highest volume stocks (IEX’s own data in 605 shows an average spread of 2.2 cents for all marketable orders sent to them), nor are they representative of the high frequency trading that are a material part of these statistics. Thus, IEX’s poor fill rates on displayed orders could simply be an example where the speed bump does matter, but not in a good way.
In addition, while in the cases described in the note, IEX was at the NBBO, their quote quality was much lower overall than the exchanges they compete with. In June, despite it being “IEX’s record displayed market share month”, IEX still lagged their competition in both their participation at the NBBO and in setting the NBBO. According to data from MayStreet, for stocks trading over 3 million shares per day, IEX was at the NBBO 36.33% of the time, while EdgA was there 47.91%, BX was there 50.67%, BatsY was at 53.33%, Nasdaq was at 91.22% and NYSE (for those stocks it trades) was at 94.83%. In fact, IEX had the lowest participation rate at the NBBO of all exchanges except for the CHX (at 18.01%). IEX’s performance in setting the NBBO was much worse. Also, according to MayStreet data, in June IEX, for stocks trading over 3 million shares per day, only set the NBBO 0.45% of the time. That compares to over 1% from BX, 1.19% from EdgA, 1.62% from BatsY, 3.3% from PSX, 25.38% from Nasdaq and (for stocks that it traded) 27.94% from NYSE. This data is quite relevant, since brokers who program routers need to be conscious of the fact that they may have latency. Thus, it makes sense, on the margin, to route to venues that are more likely to be at the NBBO in general, to compensate for the potential that the quotes they “see” might no longer be available.
As a closing note, I want to point out that it is an act of entitlement (and hubris) for IEX to chastise the industry for not routing to their displayed quotes based on their own self-interested analysis. IEX competes directly with those same brokers (by offering algorithms masquerading as order types as well as smart order routing) and have spent the last 3 years maligning brokers for their behavior. Considering that, in a situation where brokers have a choice how to meet their best execution obligations (not if, since there is zero proof that brokers are receiving lower fill rates by their choices), why would anyone be surprised that some would choose to route elsewhere? When one considers the pattern of IEX publishing out-of-context data without disclaimers, it is even harder to understand why anyone would listen to their arguments…
NYSE puts market structure arguments in their proper perspective
Very well-articulated promotional piece by the NYSE. Too many commenters, in their zeal to promote their own favorite market have lost sight of the primary function of stock exchanges that Michael Blaugrund phrases well: “Stock exchanges play a critical role in serving as the mechanism to aggregate capital in order to fund ideas, grow businesses, hire workers, and contribute to economic growth”. That is why wild accusations, out of context statistics, undisclosed self-interests, and baseless rhetoric against our capital markets needs to be fought with such diligence. The US equity market is the envy of the world and a prime source of competitive advantage for our companies. Kudos to the NYSE in taking the “high road” by solely focusing on the positive aspects of their efforts.
Bloomberg wrote a piece on how lending ETFs can offset fees to make them essentially free to own. While the data makes it look that way, in practice it is often not the case. That is because of the opaque way the stock loan market is priced. You see, retail investors who own stocks and ETFs rarely, if ever, receive the stock loan income from those holdings. (Their retail broker keeps that income as part of their terms of service). Institutions and pension funds, meanwhile, when they loan our their holdings, pay a handsome spread to the stock loan desks that facilitate those trades. I am unsure what the net return to those funds and investors that hold ETFs that are “in demand” by short sellers is, but it is far less than the quoted rates in the article. The article also makes it seem like there is “free money” to be made by loaning out ETFs, but that is also wrong. There is little evidence that there is unsatisfied loan demand for ETFs (ETNs like TVIX, probably, but that is a different story). Thus, it is likely scenario that increased supply of ETF stock for loan would mean lower rates for those stocks that are in high demand, and a lower percentage of “general collateral” stock being borrowed.