A study of TSX Alpha Exchanges speed bump made headlines last week.i Coincidentally, the study was released the same week that RBC hosted a talk with Brad Katsuyama of IEX, recently approved by the SEC as an exchange despite virulent opposition to its speed bump. I had an opportunity to ask Brad about how IEX intended to maintain its competitive advantage, given the expected proliferation of speed bumps now that the SEC has given its blessing. Brad rightly pointed out that not all speed bumps are created equal. The devil is in the details, as the TSX Alpha study reveals.
The study presented 3 key effects of Alphas speed bump combined with its inverted maker-taker model:
1. Alpha has succeeded in attracting retail order flow from brokers who are paid for taking liquidity.
2. Alpha has succeeded in attracting HFTs to provide liquidity because the speed bump protects them against adverse selection from informed or institutional order flow. The speed bump applies to all orders except post-only orders, typically used by market makers. If HFTs detect that a large order is being sprayed across exchanges, they have the time to cancel a quote before a leg of that order reaches Alpha where it will be slowed by the speed bump.
3. Effective spreads on several other Canadian exchanges widened following the relaunch of Alpha in September 2015.ii The study attributes this partly to increased segmentation between retail and institutional order flow. Institutions can bypass Alpha as an unprotected market, and have done so because they have experienced fading quotes by market makers who avoid interacting with informed flow. This has resulted in increased concentration of retail orders on Alpha and institutional orders on other exchanges.
The TSX has taken issue with the methodology of the study. For one, the study relies on proxy measures to identify retail, institutional and proprietary flow. Hence, it can only estimate what is really happening and why. Secondly, public comments by the studys chief author suggest that the intent of Alpha was simply to facilitate HFTs which he characterizes as sharks-surprisingly colourful language from an objective academic.iii
Why did the TSX reconfigure Alpha? Alphas marketing literature states that the changes were made to improve the economics of executing retail orders in Canada, particularly as an alternative to selling those orders to U.S. wholesalersiv -retail brokers increased this practice following the limits that IIROC imposed on retail order routing to dark pools in 2012.v Last year, IIROC put the kibosh on the crossborder sale of order flow by requiring brokers to demonstrate meaningful price improvement on an order-by-order basis.vi
TSX Alpha sought to attract this newly captive retail flow by paying brokers to direct marketable orders to its exchange, and providing certainty and immediacy of execution with larger displayed quotes-the
post-only orders by market makers must be a certain minimum size and have a minimum resting time. The study does not dispute that Alpha has delivered on these benefits for retail flow.
However, according to the study, Alpha has not delivered price improvement. Alpha expected that HFTs would provide better pricing than the prevailing market because they are protected against adverse selection from informed order flow. This has not occurred for two reasons, according to the study:
1. HFTs seek to recoup the payment to post liquidity with wider spreads-a commonly noted effect of inverted maker/taker models.
2. Posting aggressive orders would negate the benefit of the Alpha speed bump for HFTs. By quoting at the NBBO, HFTs buy time to gain market information from orders executed on other exchanges to determine whether toxic flow is coming their way. The last thing they want is to be hit first by providing superior quotes.
So, what, if anything, can we conclude from all this?
First, we need to study speed bumps further with more precise inputs-these typically come from regulators rather than public data. Second, we need to understand the wider market context to assess the relative advantages and disadvantages of specific innovations. For instance, are retail investors receiving better or worse pricing on public markets following the relaunch of Alpha than they did from U.S. wholesalers? We know from an IIROCsponsored study that active retail orders received inferior prices on lit markets after IIROC restricted routing of retail orders to dark pools in 2012. vii The IIROC study does not consider the resulting increase in exchange fees incurred by brokers, a pain point that Alpha alleviated by offering rebates for active flow.
Segmentation has been around since the dawn of public markets. Whether segmentation is more effective between public and private venues or among public markets is the question. As I stated in my previous post on the increasing popularity of on- and off-exchange auctions for institutions, I am highly skeptical that exchanges can effectively manage segmentation without disproportionately disadvantaging one constituency more than another. I am open to being convinced otherwise.
i Sean Foley, et al., The Value of a Millisecond: Harnessing Information in Fast, Fragmented Markets, October 2016. ii The study calculates half-spreads as the difference between the trade price and the prevailing NBBO midpoint. Realized spreads compare trade prices with the NBBO midpoint 20 seconds after the trade. Price impacts are calculated as the effective spread minus the realized spread.
iii Christina Pellegrini, TSX Speed bump doing more harm than good: study, Globe and Mail, November 7, 2016. iv TSX Alpha, Reshaping Canadas Equity Trading Landscape, October 2014. v S. Mostowfi and L. Preyma, “Canadian Equity Retail Trading: Smart Order Routing and Regulation,Tabb Group, April 2014.
vi IIROC, Re-Publication of Proposed Dark Rules Anti-Avoidance Provision, January 29, 2015. vii IIROC, Impact of Dark Rule Amendments, May 7, 2015.
