On September 8, 2015 the TMX Group published the much anticipated fee schedule for the revamped Alpha Exchange (subject to regulatory approval). Coupled with the recently published minimum size thresholds for “Post Only” orders, we now have near perfect information around how the market will operate and can opine intelligently on the expected impact the new model is likely to have on equity trading in Canada.
Before we begin to pontificate upon the migration of flow, or impact on market quality, let’s first review what the Alpha model is and how it will function. TMX is currently targeting to have these changes in place for trading on September 21, 2015. On that date, TMX will make the following alterations to their suite of marketplaces:
- Alpha will move to a “taker maker” pricing schedule – pay to post, rebate to take
liquidity. From September 21 – November 30, dealers will be charged 10 mills per share to post orders, and receive 10 mills per share when they take liquidity. As of December 1, the price of post will increase to 14 mills for regular orders, and 16 mills for Post Only orders.
- Alpha will introduce a delay (“speed bump”) for inbound orders and cancellations. The delay will be random (between 1 and 3 milliseconds).
- Alpha will allow Post Only orders to bypass the speed bump, both for order entry
and order cancellation.
- Introduce a minimum size for Post Only orders.
- Shut down TMX Select, the current taker maker venue.
- Kill off Alpha’s IntraSpread dark pool, and end the Market on Open auction
At the same time as these changes are implemented, the Ontario Securities Commission (OSC) will deem Alpha to be the first unprotected lit venue in Canada
(i.e. orders resting on Alpha will not be protected from trade through on other venues). They will then study the idea of having all venues with speed bumps
become unprotected markets.
Why is the TMX introducing this market structure?
The changes to Alpha are meant to replicate as closely as possible, within the confines of Canadian market regulation, the economics and structure of U.S.
bilateral wholesale arrangements. By changing the market to taker maker pricing, those participants taking liquidity will get paid to do so, much like U.S. retail firms do via payment for order flow arrangements. The speed bump is designed in a manner that allows liquidity providers to avoid larger, typically institutional flows and only participate against the smaller, mostly retail order set. Let’s explain this with an example:
Who benefits from this market structure?
The speed bump is designed in a manner that allows HFTs to avoid larger, typically institutional flows and only participate against the smaller mostly retail order set. Let’s explain this with a quick example.
Consider stock XYZ, currently quoted at 10.00 – 10.01. There are 5,000 shares bid on each of TSX, Chi-X and the new Alpha. ITG has an order to sell 15,000 shares. If we use our low latency – high speed router and spray 5,000 to each of the three markets the TMX and Chi-X orders execute almost instantly, but the liquidity provider posting on Alpha now has somewhere between a few hundred micro seconds and just shy of 3 milliseconds to cancel her bid and avoid trading with our inbound large order. By design, the liquidity provider is able to selectively trade against smaller, less impactful orders. The combination of taker maker pricing and the speed bump is basically a mechanism to pay for the right to trade against the most lucrative flow
and avoid trading against the most informed flow.
Even if firms were to slow down their flow, in an attempt to stagger orders, the random nature of the speed bump will make it impossible to send the orders with the necessary time precision to hit each market at the same time.
While the mechanism provides improved economics for retail dealers, and allows participants to compete for the ability to provide liquidity to the smallest orders, it comes at the expense of institutional clients. Furthermore it will negatively impact the perception of the Canadian market, as Smart Order Router fill rates fall across the board. Allowing proprietary traders to pay for the right to trade against the least informed flow and avoid larger clients can only make it tougher for natural participants to capture the passive side of the spread as they are crowded out by these strategies.
To compensate for the decreased certainty of the liquidity that will be residing on Alpha, the regulators are going to remove the protected status from the market. While some will argue that this is a reasonable fix, in reality it does nothing to compensate for the missed liquidity. Explaining to a buyside trader or portfolio manager that the 5,000 or 20,000 shares they missed weren’t protected doesn’t in any way change the fact that they were unable to capture posted liquidity. The very real takeaway for most market participants will be that order fill rates in Canada will have fallen.
How does the new model impact trade flows?
The Alpha model is clearly designed to take flow from Chi-X Canada’s CX2 venue. After the initial 10 week low fee introduction, the long term taker maker pricing model is very similar to the fee schedule currently at play on CX2. The key difference is the ability for a liquidity provider to pay an extra 2 mills to post passive orders that bypass the speed bump. This in turn allows the liquidity provider to fade her quote when facing evidence of a large liquidity taking order entering the market. Were this advantage to be priced identically to CX2 it would clearly represent a superior value to the liquidity provider. As it is, the market making community will need to determine if the ability to fade quotes is indeed worth more than the 2 mill asking price. Based on our discussions with some of the largest HFT firms in Canada we believe the price only works if Alpha is able to achieve first look on a large enough slice of the active retail order flow. If instead CX2, which has a similar take rebate, is able to keep first look, then liquidity providers will be less inclined to post on the more expensive Alpha market. The key to success will be incenting liquidity providers to post an aggregate size that is greater than available at CX2, during the first few weeks of the new model, such that retail brokers will feel they are gaining a size advantage when taking from Alpha rather than CX2.
The key to ensuring that liquidity providers post in greater size on Alpha than CX2 over the key first few weeks is a controversial introductory pricing mechanism that makes it cheaper for liquidity providers to post liquidity, but keeps the take rebate in line with CX2. As such, the TMX has proposed a schedule in which the posting fee and taking rebate are identical. Such a schedule is not without risks. Given that there are costs associated with running a matching engine, over and above any liquidity rebates, the TMX is effectively running the Alpha matching engine at a loss during the introductory period. While this may gain regulatory approval if suggested by a fledgling new entrant, the TMX is the incumbent market in Canada with the majority of cash equity market share. In Canada, the Competition Act very clearly states in Section 78, Paragraph 78(1)(i) that incumbents cannot sell “at a price lower than the
acquisition cost for the purpose of disciplining or eliminating a competitor.” These rules are in place to ensure that dominant players don’t use temporary loss leader pricing to thwart would-be competitors. (Note the author of this note is neither a lawyer, and the analysis above does not, in any way, constitute professional advice).
Beyond the potential competitive issues with the pricing model is the issue of explicitly pricing speed advantages. By charging more for Post Only orders – whose only real advantage is the ability to bypass the speed bump – the TMX is effectively creating two-tiered access to the matching engine. This would seem to violate at least the spirit, if not the prescriptive language, of the Canadian fair access rules. To be clear, unlike the Aequitas Neo book model where exposure to a speed bump is determined solely by participant type, in this model passive participants will be able to pay a higher fee to avoid the speed bump. The ability to buy your way past the speed bump will almost certainly result in pushback from dealers and buyside firms. It is very likely that dealers will forego beneficial economics on an order by order basis to avoid supporting a market model that facilitates excessive intermediation and introduces unsavory latency economics.
At the end of the day, we predict that the Alpha model will struggle to gain traction, as many of the dealers we have talked to are not in favour of it. We believe the pricing difference between Alpha and CX2 would have to be far larger to attract flow, given the controversial attributes and concern that if Alpha is successful it will breed further speed bumps and tiered latency economics from other venues.
While our prediction for Alpha is muted at best, we have done significant work to make changes to the ITG Algorithms® suite and ITG Smart Router™ in the event we are wrong, and are forced to deal with lit liquidity that is less accessible. We would welcome the ability to discuss these with clients in person, to fully explain how we believe we can best capture liquidity both passively and on the active side.
As always, these issues are complex and multifaceted. The market structure that works best for one participant may be far less favorable to another. We invite feedback and debate on our views, and look forward to a continued conversation around this and many other issues impacting our market.
Doug ClarkContributors Managing Director, Head of Research, Canada