As a futures exchange proposes a new speed bump mechanism, a number of market participants are coming out in opposition to it. Some of the arguments they’re making will sound familiar to those in the FX markets, says Galen Stops.
On February 1 the Intercontinental Exchange (ICE) put the cat amongst the proverbial pigeons by submitting an amendment to its rules that would introduce a three millisecond delay, otherwise known as a speed bump, for incoming orders that would otherwise transact immediately opposite resting or “passive” orders.
The rationale from ICE for introducing this is to “level the playing field” by reducing the speed advantage that some market participants currently enjoy, and the exchange operator said that it plans to implement this new rule for gold daily and silver daily contracts that it offers. By levelling the playing field, the exchange hopes to attract more market participants to trade on these contracts and thus foster greater liquidity.
On the surface, this might not sound like a big deal. After all, lots of platforms in various asset classes have introduced speed bumps and the contracts in question are not widely traded, so it’s not as if many firms will be impacted by the rule amendment.
And yet it is proving to be a significant point of contention, judging by the comments submitted to the Commodity Futures Trading Commission (CFTC) and made to the press following ICE’s announcement. Specifically, what is controversial about this amendment is that the proposed speed bump is asymmetrical (i.e., it only applies to one side of a potential trade). Broadly speaking, there are five key arguments being made about why such an asymmetrical application of speed bumps is detrimental for the futures markets.
1. The “Last Look” Argument
In comment letters to the CFTC certain firms objected to the asymmetrical speed bump on the grounds that it offers market makers an opportunity to modify their exchange orders when there is a price change in a related market.
“In effect, this operates like a “last look” in which market makers have a chance to selectively move their quotes out of the way when it is in their interest to do so, but to remain firm for their quotes when they already know that is to their advantage (and generally to their delayed counterparty’s disadvantage),” says the Futures Industry Association Principal Traders Group (FIA PTG) in its comment letter.
David Fox, a principal at LaSalle Asset Management, puts it somewhat more colourfully in his comments: “I will probably be one of the few market participants to comment on this who traded back when chalk boards were still in use on the exchange floors, but I have one question concerning this proposal which references the pit trading days: Isn’t this proposal akin to allowing a market maker to look at a floor broker’s order before the broker actually sees the order?”
He concludes: “I’m sure that back in the day had a group of market makers approached the CFTC with such an idea, it wouldn’t have even made it through the fax machine. Times have changed no doubt, but I was not aware that the measure of ethical standards changes with markets and times.”
Stephen Berger, managing director, global head of government and regulatory policy at Citadel Securities, and Mary Morrison, who works in compliance at Freepoint Commodities, likewise echo these comments and specifically refer to the practice of “last look”, which both state “has been denounced in other markets”. Presumably one of the markets they are referring to here is FX, where the functionality of last look has generated no small amount of controversy in recent years.
Indeed, a source at one trading firm says that last look functionality is even more dangerous in the futures market than in FX because the former lacks some of the self policing mechanisms of the latter. Explaining this comment, they point out that in FX there is a greater ability to create custom liquidity pools, whereby if a liquidity provider (LP) is rejecting too many trades using last look, the liquidity consumer has the option to remove them from the pool of liquidity that they see.
And The Counter Argument
Yet, while superficially the comparison to the last look functionality that exists in FX seems appropriate, even a cursory reading of the actual proposal from ICE highlights a crucial difference.
Here’s how the asymmetrical speed bump actually works: when the exchange receives an incoming order that would match a passive order resting on the central limit order book (CLOB), the execution of this trade will be delayed by three milliseconds, a window during which the firm that placed the resting order may modify it.
Now here’s the important line from the original submission: “The trader who submitted the resting order will not be notified that there is an incoming order waiting to take liquidity”, although they will have “time to react to external market conditions (e.g., a price change in the related market)”.
This is the fundamental difference – in FX the market maker is able to see the order before they decide whether or not to actually trade on it, whereas in the ICE proposal the market maker cannot see the incoming orders and therefore any decision to modify their price will not be dictated by these orders.
Or, as XTX Markets explains in its comment letter to the CFTC: “Asymmetric speed bumps enable market makers to post better prices at larger size because traders engaged in latency arbitrage are unable to deploy that strategy on such a market. Natural liquidity consumers continue to experience high fill rates because their consumption of liquidity is not driven by millisecond level external events. lmportantly, market makers benefiting from these passive order protections never have knowledge of an incoming order that is in the delay mechanism. This fact is a material differentiator from what is commonly referred to as ‘last look’ in the FX markets, where a market maker is presented with an order and has some period to decide whether to accept it (or potentially engage in abusive activity such as trading ahead of that order).”
Similarly, Steve Crutchfield, head of market structure at the Chicago Trading Company (CTC), points out in his comment letter that the ICE proposal entails zero information leakage and, because the market maker cannot see the incoming order, “is therefore, by definition, not a ‘last look’ in any sense of the word”.
…….(And the Counter-Counter Arguments)
Now, the counter-argument to this given by Berger is that an LP does not need to actually see the incoming order to benefit from this functionality. Instead, he argues that the LP only has to assess current market data in real-time to determine whether the market has moved against them while potential incoming orders are being delayed.
“In practice, this ‘last look’ allows liquidity providers to honour their quotations only when it is in their ultimate economic interest, and to otherwise back away from their quotations at the expense of other market participants,” he adds.
But Crutchfield scorns the idea that the proposed speed bump would hurt end investors, stating that “the probability is effectively zero” that an actual end user would, for unrelated reasons, send an order on the relevant side of the market during the precise millisecond-level window corresponding to a contemporaneous market-moving event that causes a liquidity provider to reprice.
He adds: “Aside from such extraordinary coincidences of timing, the only orders that would generally be deprived of executions by the Proposal are those sent by sophisticated electronic trading firms engaging in high-speed sniping based on immediately prior receipt of market-moving information.”
Let’s say that ICE implements this speed bump and CME continues to operate without one. If the price on CME ticked in one direction both the market makers and the takers would see that update at the same time. However, a firm engaging in latency arbitrage strategies would be able use its high speed systems to take that information and pick off the market makers that can’t react as quickly. What the speed bump does, proponents of this view say, is negate the advantage that such HFTs currently enjoy.
“If you’re an asset manager trying to trade gold, you’re not sitting there with microwave towers arbitraging between two different venues and getting data from one place to the other before other people. They’re not responding to microsecond level events in the order book, so when you have a speed bump it’s bad for latency arbitrageurs because they lose an existing advantage, but it’s good for long-term investors and end users,” says one market source.
2. The Phantom Liquidity Argument
This is another argument that FX market participants should be intimately familiar with. The crux of the argument coming from futures market participants that are against the proposed ICE rule is that it will create a market where the prices that firms see on their screens cannot be relied upon.
The fact that the asymmetrical speed bump will, according to Morrison, create “the mere illusion of greater liquidity that is inaccessible”, has a number of significant consequences, these firms contend.
A source at one trading firm says that they worry that in today’s computer driven trading environment this could create feedback loops during stressed market conditions. For example, they say that if prices in gold or silver started to move on other exchanges there would be a time lag before the price changed on ICE because the firms with resting orders there wouldn’t need to update them as quickly with the speed bump in place. The consequence of this, they contend, is that computers will keep routing orders to ICE because it will appear to have the best prices available in the market. These orders will not get filled because by the time the incoming order gets past the speed bump the resting orders will have been updated in response to changing prices elsewhere. Yet the prices on ICE will continue to be better than on the other exchanges and so the computers will naturally keep sending orders there, orders that won’t be filled, and on and on the process would go.
Another gripe about phantom liquidity being made by some sources in the futures markets is that it devalues the data that they’re currently paying so much money for. The argument here is that if firms can’t rely on the data that they’re purchasing to design algos and improve the quality of their execution to the benefit of their customers then that data is being devalued and all the work that has gone into making this market deterministic is being undone.
One source explains that because the liquidity on ICE exchange would no longer be “firm”, it means that their probability of getting filled on a bid or offer on that platform has inherently been reduced from “X” to a fraction of “X” and therefore they cannot build in precise latency and cannot count on the liquidity that they see or the data that they’ve bought.
And the Counter Argument
The response to this argument once again focuses on the different trading styles of end users and firms relying on latency arbitrage strategies. Proponents of the speed bump say that, yes, some liquidity will prove to be phantom, but only to firms that are trying to use speed to pick off other market participants. For real end users, they contend, the prices available will be very real and hittable.
One source points to the speed bump employed by EBS, which is symmetrical and allows for cancellations, as an example of this. So in this example both the market maker and the market taker’s orders are delayed, if either side decides to cancel the order during the delay they can do so and if both orders are still there at the end of the delay then they’re matched. A firm that isn’t engaged in latency arbitrage will put in an order and then the price is generally good for hundreds of milliseconds, it doesn’t whip around much. Thus a counterparty trading at a similar speed will be able to hit the pricing there.
By contrast, a firm that is dealing and trading in nanoseconds would have an informational advantage without a speed bump, but with one by the time the orders are delayed both counterparties have the same view of the world and therefore their order might miss as the market maker catches up and updates their pricing.
“Arbitrageurs will get low fill rates because that’s what speed bumps do, they pick up latency arbitrage. But end users are just not trading in this way, they’re not involved in these inter-market latency races, they’re just trying to trade the market,” says the source.
The end result of this, they contend, is that market makers will be able to show better prices because they don’t have to worry about getting picked off by latency arbitrage strategies and end users will see deeper order books.
This source says that the issue of making markets less deterministic is a very separate one, but that is also a red herring, in this debate. They explain that in fact exchanges have become too deterministic, whereby if an order is placed a nanosecond before another it will invariably get filled, creating a winner takes all system where the winner is always the fastest. Their logic states that investors and end-users of financial products are not operating within a nanosecond time frame and thus will consistently lose out to the firms that are, meaning that the firms making the prices will be forced to widen out their spreads to account for this.
Technology has clearly brought efficiency and progress to markets, but while it was very useful when markets went from trading in hours to minutes to seconds, there is a case for arguing that markets could be reaching a point of diminishing returns, where being a few nanoseconds quicker means that firm wins everything, reducing competition and providing no real benefit to the end investor. In this view, determinism in and of itself is not the be all and end all.
3. The Market Manipulation Argument
Some market participants claim that the new rules being proposed by ICE could encourage market manipulation, and specifically they say that because the speed bump would give firms the ability to cancel resting orders with less risk of order execution it may increase the instances of spoofing, where LPs post quotations that they have no intention on honouring but are instead aimed at moving market prices.
Berger says that such behaviour could become most pronounced during periods of market volatility, when LPs will more likely to takes advantage of the ability to cancel or modify resting quotations.
“ICE has failed to detail how the exchange will monitor and prevent this type of trading activity from occurring as a result of the Asymmetric Delay, including how it will distinguish between bona fide cancellations and disruptive trading activity. In addition, monitoring and surveillance is not an adequate cure for improper market structure design,” says Berger in Citadel Securities’ comments to the CFTC.
Similarly, Jennifer Han, associate general counsel at the Managed Funds Association (MFA), says that the proposed ICE rule amendment could make it harder for the CFTC to bring enforcement cases against market participants guilty of market manipulation as it would have to prove that the trader in question was spoofing and not merely cancelling a resting order.
And the Counter Argument
However, CTA/CPO representatives from the National Futures Association (NFA) board of directors express doubt about these claims, pointing out that spoofing is not typically done through resting orders and therefore is unlikely to increase if the rule amendment is approved. In addition they say that this kind of market abuse is already subject to criminal sanction which acts as a material deterrent.
These representatives add: “The proposal may in fact increase liquidity (the intent by ICE) by improving the state of resting orders v the incoming orders, thereby fostering the placing of more resting orders and the creation of a deeper market.”
In addition, Trabue Bland, president of ICE Futures US, points out in his comment letter that the proposed speed bump won’t encourage spoofing and false volume because the amount of time that an order appears and is exposed in the order book is unchanged with the addition of this functionality.
4. The Level Playing Field Argument
A common theme in a number of the responses is that by introducing an asymmetrical speed bump ICE is tipping the scales in favour of LPs and that is in turn inconsistent with existing CFTC rules that designated contract markets (DCMs) – such as exchanges – are subject to.
Core Principle 2 from Section 3 of the Commodity Exchange Act (CEA) says that DCMs have to provide impartial, transparent and non-discriminatory access to its markets and services, while Principle 9 requires that they must “provide a competitive, open and efficient market and mechanism for executing transactions that protects the price discovery of trading in the centralised market” of the exchange.
Han says that the MFA is “concerned” that ICE’s new rule proposal could create an unfair market by providing certain market participants trading advantages over others, and that it could violate the impartial access requirements by applying a delay to incoming but not resting orders.
Berger is a little more strident in his comments on this issue. He argues that, by allowing a firm to selectively decide whether or not to honour firm, displayed quotations based on current market conditions, the proposed rule clearly gives LPs a structural advantage that results in three forms of unfair discrimination.
From his comment letter: “First, these liquidity providers are advantaged compared to liquidity takers when interacting on ICE, as they can opt to honour their quotations only when it is in their economic interest to do so, unlike other exchange participants. Second, since only these liquidity providers have the technological capabilities to utilise the asymmetric delay, they are advantaged compared to other market participants submitting resting orders on the exchange. Third, these liquidity providers on ICE will be advantaged compared to liquidity providers operating in other markets. Specifically, the liquidity providers on ICE will be able to leverage the price discovery being provided by liquidity providers in other markets by using the associated data regarding quotations and executions to adjust their displayed quotations on ICE to avoid unfavorable executions.”
The consequence of this, Berger argues, is that market liquidity and price discovery might be negatively impacted as LCs are denied access to firm and reliable prices, while other market participants may be discouraged from submitting resting orders if they are unable to effectively utilise the delay. He adds that price discovery on other markets may also degrade due to those liquidity providers being placed at a disadvantage compared to the LPs operating on ICE.
Privately, some sources describe the proposed speed bump as a form of payment to market makers, both in the sense that their trading will become more profitable and they won’t have to make investments in technology to compete with faster firms. They argue that it’s not their fault that other market participants haven’t embraced and built the technology needed to compete in modern markets, and that this speed bump is simply a mechanism to protect their business models.
And the Counter Argument
Now the flipside to this perspective, as alluded to previously, is the argument that the futures markets are currently not a level playing field and that ICE’s new proposal is actually seeking to redress this very problem.
The NFA board of directors point out that in the days of open outcry the floor traders clearly had an opportunity to profit from customer orders, and claim that the co-located computer has now replaced the floor, with the exchanges – for a price, of course – providing lower latency access to markets, which they contend inherently puts other market participants at a disadvantage.
“Trading has evolved to become more of a speed contest than a price competition, as increasing resources are devoted to reducing latency to gain a time advantage,” they add.
Crutchfield concurs with this assessment, claiming that right now market participants can deploy strategies that use their speed advantage to execute against resting quotes before they can be changed in response to new relevant information, such as changing prices in a related market, which is known as “sniping” or “pick-offs”.
“These changing prices in the related market also drive the party who submitted the resting quotes to simultaneously attempt to respond by repricing their quotes – indeed, that party may have a quote at a new price already “in flight” on its way to the exchange, or even pending acknowledgement inside the exchange’s matching engine, when they receive the adverse execution triggered by the “sniper”. Under current market structure, however, if the sniper is faster by any amount of time – often a millionth or a billionth of a second they can force the liquidity provider to trade with them at an immediate loss. This creates a huge incentive for some trading firms to invest in messaging speed at the expense of all else,” he says in the CFTC comment letter.
Crutchfield contends that this creates a never-ending arms speed race that is counter to investor protection and the public interest because liquidity providers who don’t want to make the ongoing investment necessary to participate in this race will be forced to widen their spreads, decrease quote sizes or potentially even stop displaying liquidity altogether.
Likewise, XTX Markets says that “the race for speed in trading has reached an inflection point” where the cost of gaining an edge over other market participants is harming liquidity consumers.
“Market makers providing liquidity need to price to the average of the toxicity of the order flow they interact with, and to the extent that they are being adversely selected by latency arbitrage strategies, they must widen their spreads, which in turn increases the costs of trading for all liquidity consumers accessing that market. Moreover, the need to compete on speed at these levels creates a meaningful barrier to new entrants into market making who may have unique pricing, as well as time horizon and risk absorption capabilities,” the firm says in its comments.
Each of these firms point out that the proposed asymmetric speed bump would mitigate the effects of this arms speed race, increasing the competition to provide liquidity and encouraging the firms quoting prices to offer tighter spreads and larger sizes. XTX even argues that the ICE proposal is not a novel one, and says that it is part of a growing trend of similar mechanisms that other platforms have introduced to protect market makers from latency arbitrage strategies, citing 12 different examples – including Eurex, EBS Market, Thomson Reuters (now Refinitv) Matching, ParFX and Moscow Exchange that have all introduced such mechanisms for their FX markets.
5. The Slippery Slope Argument
The crux of this argument is that, should the CFTC approve the ICE rule amendment, it could lead to similar asymmetric speed bumps being introduced elsewhere and, ultimately, across all the major futures exchanges.
The FIA PTG appears suspicious that this is the end-game that ICE has in mind because it notes in its comment letter that, to learn more about the proposed rule change, ICE held an open call with its membership in mid-December, during which they were assured that it was an experiment designed to foster liquidity on two otherwise illiquid contracts and that the exchange had no plans to apply this functionality to other products.
Thus the FIA PTG was left “surprised and dismayed” to find that when ICE actually submitted the proposed change to the CFTC the language used did not specify that the asymmetric speed bump would only be applied to those two contracts and said that the length of the speed bump would be left to the discretion of the exchange.
This, of course, leaves open the possibility that ICE could subsequently apply such a speed bump to any of its futures contracts, and further, that this would in turn create a precedent that could be replicated across the entire futures market. The worry here is that exchanges will end up using latency mechanisms as a means of competition, in order to go after the liquidity on other exchanges.
This argument runs that if one exchange is offering an advantage to LPs – in this case by offering them a chance to adjust their prices to the market ahead of incoming orders – that will make trading there more profitable, then they will inevitably gravitate towards trading on that exchange, forcing other exchanges to offer either the same or comparable advantages in order to counteract this.
One source says that something similar has already occurred in equities, where exchanges started to compete around different order types that could benefit different market participants in order to encourage them to trade on their platform. The ICE proposal, they add, is really differentiation pretending to be innovation by the exchange.
The slippery slope argument can also be applied to the design of the speed bumps themselves, with some market participants arguing that if ICE introduces a three millisecond speed bump then another exchange might then introduce a four millisecond one, then another a five millisecond one, and so on, and so on.
When Profit & Loss suggested to one market source making this argument that exchanges were unlikely to prevent a race to the bottom in terms of speed with a new “race to the top” with ever longer trade delays because, at a certain point, the actual impact of different delay times would become negligible, they responded that simply introducing latency is basically a form of payment to market makers.
And the Counter Argument
Well for starters, the language used by some trading firms to make this slippery slope argument is certainly hyperbolic. One source warns that the industry could be “opening Pandora’s Box” and that if the CFTC were to approve the proposed rule change it would be “risking the integrity of the entire futures market”, while another claims that before long, market participants won’t be able to rely on any of the liquidity in the futures market.
Even if one accepts the arguments that ICE’s proposed speed bump will create a last look type advantage for LPs, in FX although this practice has generated some heated debate it has hardly precipitated a collapse in the integrity of the market or caused a debilitating inability to rely on the prices there.
There is almost certainly some truth to the idea that the main motivation for ICE to introduce an asymmetrical speed bump is to grow liquidity on these metals contracts, but this is akin to accusing a for-profit exchange of seeking to grow and expand its business.
While ICE left the language in its proposed rule amendment open enough that it could apply the speed bump to other contracts, there appears to be no doubt that it is only starting with the two illiquid contracts. So once again, if you accept all the arguments made against the speed bump – namely that it will harm price discovery, liquidity, market resiliency, create unfair competition and make market manipulation easier – then surely if these predictions of deteriorating market quality are borne out once ICE introduces it then the exchange will never bother to introduce such a latency mechanism to any other contracts.
Why be concerned about ICE introducing this speed bump for two small contracts that basically no one trades if you’re confident that this will only prove that it was wrong to do so?
The response from more than one market source to this question was essentially: why wait for a bad thing to happen to the market if you can come out in advance and stop it? They are, they claim, trying to do the responsible thing in voicing their objections.
The slippery slope argument only really works, however, if you accept the premise that the asymmetrical speed bump is a bad idea in the first place. If you instead believe, as some firms clearly do, that it is a positive innovation for the futures industry that has the potential to improve the quality of the market, then this slippery slope changes from being a threat to a promise.
And in the End…
Ultimately, despite all the complexities inherent in these arguments, whether firms are in favour or against the proposed speed bump hinges on whether they think it is levelling the playing field or tipping the scales in the favour of market makers. There are, however, actually some potential options available that would offer a compromise between these two opposing views.
For example, Han says that if the CFTC is inclined to approve the proposed speed bump, “given the novel and complex issues that it raises”, it should only do on the specific metals contracts and for a pilot period of 12 months. This could seem like a fair compromise as it would allow ICE to collect data regarding liquidity and pricing that could then be used to settle the debate regarding the impact of such a speed bump one way or the other. It’s worth pointing out though that one year of data may do no such thing – after all, the TSX Alpha cash equities exchange in Canada, which deploys a randomised 1-3 millisecond delay on all orders other than passive post-only orders, is held up by two separate comment letters as an example why the CFTC both should and shouldn’t approve the ICE speed bump.
Alternatively, ICE could implement a symmetrical speed bump that would delay the orders on both sides. Ostensibly, the asymmetrical speed bump is being used to level the playing field and not to give one side an advantage, while the objections to it are ostensibly not against speed bumps per se but the fact that only one side of a trade is subject to it.
A source at ICE claims that a symmetrical speed bump wouldn’t solve the underlying issue, which is to protect against passive liquidity. They say that if all orders are delayed the fastest side still wins, which is why the exchange is delaying orders that will execute against the resting book.
Perhaps unsurprisingly at this point, two different sources at two different trading firms offer very different responses to this claim.
One says that a symmetrical speed bump would do just as well as an asymmetrical one, adding that a mechanism like that employed by EBS, which as referenced previously has a symmetrical speed bump with both sides able to cancel during the delay, would “be absolutely fine in my book”.
The other source says that speed bumps in general don’t actually blunt the latency arms race, they just change where it is taking place. All that would happen, they say, is that firms will be rushing to cancel orders quicker than other market participants.
Okay, maybe there isn’t a compromise that will make everyone happy in this scenario, and if the CFTC has to choose between allowing the proposed ICE speed bump or blocking it, the agency should lean towards the former. On balance, the probability that such a latency mechanism could improve the futures markets by negating the speed advantage of some players seems higher than the probability that it will cause the quality of the market to deteriorate in the silver and gold contracts.