One of the biggest financial stories of 2014 was the rise of John Connor exchanges designed to combat the supposed threat of high-frequency traders (HFTs). These exchanges all operate using a specific technique – a "speed bump" that delays order fulfilment in order to combat HFT's speed advantage. The first group to employ that strategy was New York-based, Canadian-run IEX, which imposed a small delay on all orders. This fall, the Ontario Securities Commission approved of Aequitas Innovation's NEO Exchange whose speed bump is specifically targeted at high-frequency traders. Now, the TMX Group is getting into the game with its Alpha exchange through a series of modifications that the OSC is seeking comment on.
Well, sort of. The modifications to Alpha are first aimed at solving a different problem – Canadian retail order flow being diverted to U.S. markets where wholesale exchanges pay brokers for traffic.
Alpha's proposal has three elements. The first is a randomized speed bump of up to 25 milliseconds (which is, incredibly, an eternity in stock market time) that applies to all trades, HFT or otherwise. The second element is an exemption from the speed bump for large "post only" orders. That is, large standing orders that are not immediately executable at the current market price and therefore provide the market with liquidity. The third is a "taker-maker" fee that induces brokers to trade on Alpha by charging those who post these large, liquidity-generating orders a small fee and giving a rebate to brokers who actively trade on the market.
Note that this is not a wholesale anti-HFT strategy. It recognizes that HFTs are not Skynet and actually do some good: HFTs provide liquidity and Alpha gives HFTs who provide liquidity through large open orders a speed advantage over those that do not. In theory, this is a boon for those HFTs – because of the speed bump, they can observe large orders, remove ("fade" in finance jargon) some of their post-only orders, and adjust prices accordingly. This should encourage HFTs and other liquidity providers to post larger orders on Alpha. Investors would be compensated for this through the taker-maker fee, with the rebate the broker receives, in theory, being passed along to investors. The end result would be, again, in theory, that retail investors are provided a large, liquid market where they transact with HFTs but are compensated for any increase in price due to order fading through the rebate system.
Are we good? Good, because now things get hairy. Alpha is what is called a "protected exchange", one that is subject to something called the "order protection rule". I've discussed the OPR before with respect to Aequitas but, in essence, the OPR requires that exchanges route orders to other exchanges displaying a lower price before filling orders at an inferior price. For example, say there is an order for 10,000 shares of a stock at $1.00, and the TSX has 5,000 for sale at $1.00 and 5,000 for sale at $1.01, while Alpha has 2,500 shares for sale at $1.00. The OPR directs that 5,000 shares be bought on the TSX at $1.00 and 2,500 be bough on Alpha for $1.00, before the rest of the order can be filled at $1.01 on the TSX.
Immediate execution of these large orders is important for institutional investors who often exhaust liquidity on a given market. Moreover, institutional investors want to avoid so-called "information leakage," which is a fancy way of saying that they want to avoid showing sellers that there is large demand for the stock they are interested in buying. If sellers observe large demand they will do what basic economics tells them to do – they raise the price. Hence the HFT technique of using post orders to suss out demand (or supply as the case may be) only to fade the order and come back with a higher (or lower) price. While this lowers bid-ask spreads, it's an immense frustration to large investors. To combat this problem, institutional investors "spray" orders across multiple exchanges in order to realize the best possible price by executing all orders simultaneously at the best possible price.
Alpha's status as a protected market, along with the speed bump, complicates this strategy. Alpha's speed bump gives HFTs who post orders on multiple exchanges – or only on Alpha – a chance to observe marketplace dynamics and adjust prices. So, in my initial example, an order placed with both the TSX and Alpha at the same time will be subject to Alpha's speed bump. Since no trade can be executed at $1.01 before all the trades at $1.00 have cleared, the speed bump will give HFTs the opportunity to observe demand, fade their order, and come back with a higher price. The supposed liquidity will be gone at the desired price and large orders will execute at a higher price.
Since the OPR means that orders must go through Alpha no matter what, the other option is to route orders through Alpha first. In that case, the delay would allow HFTs (who often post orders on multiple exchanges) to pull the resting $1.00 order from the TSX (and the random nature of the speed bump means that investors can't co-ordinate its orders), forcing investors to execute at a higher price.
In both situations, the displayed price under the OPR will be better than the actual price that the order executes at, all because the OPR turns the speed bump into an externality – the costs imposed on investors because of the Alpha speed bump are transferred to investors attempting to execute orders across all exchanges. In effect, the OPR would make Alpha's speed bump a feature of all protected exchanges.
One solution to this problem would be for the OSC to approve of Alpha's changes but to revoke its status as a protected exchange. Alpha would then become a venue where HFTs could post large liquidity and trade almost exclusively with retail investors with small orders. HFTs would benefit from the speed bump (while still retaining a speed advantage in the posting of large orders) and the reduced risk in engaging with relatively uninformed retail investors as opposed to taking the opposite side of trades with relatively smart institutional money. Retail investors would benefit from improved liquidity and rebates.
However, a partial-OPR environment with protected and unprotected exchanges would increase complexity, make compliance more difficult and hurt the price discovery process as information would be communicated more slowly across markets. While it may divert orders back to Canada from the U.S., the benefits of that improved traffic would not necessarily be shared across exchanges.
There is a better solution. If regulators want to foster competition between exchanges, the OPR should be replaced with a strengthened "best execution" regime requiring brokers to get the best deal possible when executing an order on behalf of customers, this would allow brokers to factor in both rebates and ease of execution in addition to price when executing orders. This would allow upstart exchanges to compete on a variety of dimensions including liquidity and fee structure.
While others have argued that HFTs benefit retail investors, Alpha's proposal seems particularly stark insofar as it harms large institutional orders while providing retail investors with large liquidity. On its own, Alpha may very well be a popular exchange; as a protected exchange, it seems that any liquidity benefits that Alpha provides may be offset by increased costs to institutional investors. It's complicated stuff, but what to do about the OPR in the context of innovative exchanges is probably the most important Canadian market structure question of 2015. And just like Arnie, the OSC will be back to answer it in the New Year.