IEX continues to obfuscate…
IEX wont accept my challenge of a public debate , but they continue to write about my analysis and attempt to paper over their flaws. The latest missive, from Elaine Wah, starts with a flawed premise. Her argument, that since IEX receives midpoint orders before receiving a marketable order often, it “proves” that its OK to conflate the two.
Really??? ALL that proves is that IEX (deservedly) is a favorite location for sequential routers which probe the midpoint first, then cross the spread. Such routers typically go dark pools first and then to displayed venues, meaning, quite clearly, their own data PROVES that the routing world treats them as a dark pool…
None of this changes the fact that to compare IEX to other markets, that the orders should be separated; if anything, it makes it MORE obvious since it shows that the orders are distinct! In essence Ms. Wah makes a circular argument:
Because we (IEX) are routed to like a dark pool, you should group all of our orders together when comparing us to an exchange, even though they are treated differently.
The key is that the analysis of which market is best to route to, should be driven by the type of order routed, and not the type of order that was resting on the exchange.
A word on Displayed vs Non Displayed (Dark) Orders
Ms. Wah also decided to confuse the differences between DISPLAYED and DARK orders with the comparison of marketable and non-marketable liquidity seeking orders. This would be amusing, if it was not part of a fact pattern of obfuscation, where IEX continues to claim that their speed bump helps resting displayed orders. (It doesn’t, since the speed bump does not change the relative timing of orders processed by their price-time matching engine).
I assume she wrote this because I suggested that it was more valid to categorize by routing rationale and that there is a difference between routing to a venue because they displayed a quote or not. My point was that midpoint probing is clearly not related to displayed quotes, while marketable routing could be, and, therefore, is the only type to consider when evaluating displayed venues. I stand by that analysis, but do concede that it would be better to analyze all venues from the perspective of routing rationale, but we do not have public data for that. I will point out that IEXs router does show that their market offers more price improvement on the mix of orders they route. That is worth serious consideration by routing firms, but those firms also need to include other dark pools in that analysis, and also fill probability in various scenarios.
If the displayed vs dark confusion was purely part of a debate with IEX, I might be tempted to let it go, but the core concept behind enhanced public routing and market center disclosures is at stake. To provide value to the investing public, the statistics must group orders in a way that facilitates clean comparisons across brokers and markets with different business models. In practice, this means separation by marketable vs non marketable orders as a key first step. Next, displayed and dark should be separated as they have very different characteristics. (as an aside, both US, in the tick pilot data collection, and the European Regulators, in MiFID II, got this right, and required this separation in new reporting)