This is the full transcript of the court hearing between plaintiff Citadel LLC and the US Securities and Exchange Commission (SEC) at the United States Court of Appeals for the District of Columbia Circuit.
This case was filed by Citadel securities suing the SEC for implementing a new stock order type called D-Limit; A way to give traders a way to buy and sell stocks at the IEX while shielding them against volatile/manipulative price moves.
Unknown Speaker 0:00
Case Number 20-1424 Citadel securities, LLC Petitioner versus Securities and Exchange Commission, Mr. Wall for the petitioner Miss Berisi for the respondent, Miss Stetson for the intervener for the respondent.
Judge Rao 0:15
Good morning, Mr. Wall whenever you’re ready.
Wall (Citadel Lawyer) 0:18
Good morning, Your Honours, may it please the Court, the SEC has allowed a small for profit exchange to interfere directly in the market in order to attract one side of that market and boost its revenues. In approving that change. The Commission committed three errors that independently require reversal. First, the Commission seriously misunderstood the data before it. Rather than look at particular trades. The Commission relied on misleading statistics that show a lot of trading when the signal is on. Those statistics reflect normal trading activity, not phantom arbitrage. Large orders across multiple exchanges cause prices to move, which turns on the signal, even after Citadel submitted data showing exactly that a Joint Appendix 372 The Commission never addressed it. Second, even
Judge Walker 1:07
Mr. Wall, you just referred to phantom arbitrage. Do you do you think latency arbitrage exists?
Wall (Citadel Lawyer) 1:15
Judge Walker, I don’t think the court has to get into it all the court needs say is that if it exists, the data here doesn’t show it?
Judge Walker 1:23
Well, but But part of the argument is whether or not the algorithm and the speed bump and you know, all of all the things that iEX have done here is sufficiently tailored enough to reduce latency arbitrage. And I couldn’t tell from your brief whether your client acknowledges its existence even and if you don’t acknowledge its existence, it’s hard to know how you can make it an argument that the algorithm and the speed bump are not sufficiently tailored
Wall (Citadel Lawyer) 1:54
Judge Walker, I think there’s good reason to be extremely sceptical 10, 12 years ago. Sure, maybe; now? No. Look, there are a lot of dogs that ought to be barking, if this is the phenomenon that they say it is. The SEC has all the trading data in its might a system and iEX has access to the trades too. You’d think that somebody would be analysing the actual trades to give you examples of latency arbitrage. Or you at least have firsthand accounts of people who were saying that particular cables were shorter, or here are the firms that have the better data speeds. In fact, the A Miki in this case like x dx on the other side, which is one of the citadels competitors has the same connectivity and data I doubt Miss brazier. Mr. Watson will tell you that they have any competitive advantage or disadvantage visa V. Citadel. So I think there are good reasons to be sceptical Judge Walker about whether this narrative in this day and age is true. But again, I don’t need the court to agree with me on that. All the court needs to say is, look, they took there to be a problem because they took these numbers and said there’s a lot of trading and the signal is on. And the numbers are meaningless, because it’s large trades that are turning on the signal. And that’s the example that we gave at page 372 of the Joint Appendix. If you look, we supplied an actual trade to iEX and the Commission and said, We sent out a large buy order across multiple exchanges. Because we were sweeping the market, we wanted a lot of shares of this particular stock. And IEX because the speed bump worked at what was happening on all the other exchanges correctly predicted that because there was a spike in demand prices, were going to move, it turned on the signal. And then you can’t execute the last little piece and IEX. And the key if
Judge Walker 3:33
you can finish your sentence, please.
Wall (Citadel Lawyer) 3:35
Oh, I was just gonna say Judge Walker, the key thing to understand is IEX is a bit player, it only has about 2% market share, it doesn’t have a lot of liquidity there so to speak, there’s not a lot of stuff sitting on its shelves. So if you’re going you’re often going because you need a lot of something including the last little piece at a place like iEX. And it’s that last little piece that’s making the signal turn on. It’s got nothing to do with latency arbitrage.
Judge Walker 3:57
What Let me ask iEX says in its brief that that you carefully avoided telling the commission in this court two things and so I wanted to give you the opportunity to to answer them if you want to or to confirm that that sorted out would would rather not. The first is this citadel carefully avoided telling the commission this court, how much of its trading activity when the indicator is on is on behalf of retail investors.
Wall (Citadel Lawyer) 4:26
So not true, as we know, however,
Judge Walker 4:29
how much how much is how much of your trading activity is on behalf of retail investors when the indicator is on.
Wall (Citadel Lawyer) 4:37
So we looked at one month of data and as we told them at page 313 of the Joint Appendix, we said more than 50% of the trades we sent you in one month. It’s actually 56% were retail rather than proprietary. And of those Judge Walker we said 15% turned on the signal because they were sufficiently large. It was the size of the order that was driving I mean, this had nothing to do with latency arbitrage. In response to that, here’s what the commission did. It said, Okay, we you say we mislabeled you as proprietary, when in fact you do retail, we’re just going to exclude your data from the you know,
Judge Walker 5:13
I caught that I remember that part. Let me ask. The second thing that they say you avoided telling the Commission, the court, they say you did not say whether retail investors as opposed to Citadel benefit from routing retail orders, at the moment, the indicator is on.
Wall (Citadel Lawyer) 5:29
So, not true Judge Walker. But it’s the flip side, as we explain to the Commission, the retail investor is harmed, we’re not bundling these retail orders, there’s some anonymous comment letter, it provides no support for that. We told the commission it wasn’t true. And it’s not true. A retail order comes in, maybe we fill it internally, if we have the volume, or we send it out to the exchange, we send it out quickly, we send it out promptly, and we do it across all the exchanges. If it’s a large order, again, we gave an example of page 372 of the Joint Appendix. And what happens is, let me give you a really simple example. Just to make it concrete, let’s say you want 1000 shares of Apple, there’s 500 on the New York Stock Exchange, there’s 400 on NASDAQ, and there’s 100 and iEX. Right? You hit the New York Stock Exchange and NASDAQ you get your 900 chairs, IEX sees that going on it correctly predicts that because of the spike in demand, the price is going to move, it takes up a cent. So for those last 100 shares, that retail investor you wanted to 1000 shares your mutual fund or you’re investing on behalf of ordinary folks, you pay an extra dollar, their only answer that is not to look at the retail cost, they never consider that at all. All they say in the order and then in their brief is well you can just reroute. But the problem does, Walker is now you jump from the frying pan into the fire. If you executed iEX. First for the 100 chairs is true, you don’t pay the extra dollar. But you hold back. That’s the other two orders for the 900 chairs. And as we told the commission, by the time you send them out, the shares may not be there, or they may not be there at that price. So you’re sort of damned if you do even more damned if you don’t. And if the commission had looked at all of this in some reasoned order and said, Here are the benefits and the costs, I’d granted be a tougher case. But essentially, they haven’t looked at any of it. They say at page 41 of their brief, you can just reroute. We come back and say well, you said in response to the CBOE proposal, the routing wasn’t enough. And then they dropped this footnote at page 55. To say, Well, you’re not going to need any material changes in routing, the brief is totally schizophrenic on it, you will need to reroute that rerouting has costs and the Commission didn’t consider those costs.
Judge Rao 7:39
Mr. Hall, is your argument that the I mean, are you making an argument that the SEC lacks the authority to counteract latency arbitrage. So if they had provided more reasoning or more data that they could attack this problem? Or is your argument that they just simply don’t have the authority under their statute?
Wall (Citadel Lawyer) 8:01
No, only on the only on the second of the protected quote side, they can’t do what they’re do. They can’t protect the courts, because it’s not consistent with the language of the regulation. But on the first of the two issues, the D limit argument, no judge route, if they had come in and presented actual evidence of latency arbitrage, they presented actual trades, or even firsthand accounts from people who were writing the algorithms involved, if they had some reason to basis. And then if they had shown, look, we’ve isolated out the retail, we’ve isolated out the ISOs, the inner market sweeps, we’re still showing that we’re actually picking up the arbitrage. We know that’s going to impose some costs on retail investors on the liquidity taker side of the market. But here’s why we think those costs are over balanced by benefits. I mean, if they’ve done what you would expect in a well reasoned to commission order, I think it’s possible they could have done this. But this is just a sloppy order. It takes data it misreads the stats, it assumes away all the costs. If you look at both the order and their brief to this court, it reads like it’s all roses and no thorns. It doesn’t have any consideration of the costs on retail or routing. And then just to finish up where I started, Judge Rao, at the at a minimum, this and this was a third error I was going to get to in my opening, they can’t protect the quotes, the regulation says immediately and automatically executable. And the whole point of this is they’re slowing it down so that it’s not immediate. And then instead of automatically executing, they’re automatically repricing, they’re doing the exact opposite of what the regulation says. And so no on that side of it, they can’t do this absent another notice and comment rule-making
Judge Rao 9:38
So Mr Wall, on the question of immediately, you know, the Commission claims that it should get Kaiser deference for its interpretation of immediately and it made that interpretation in 2016 when they did the when they did the initial iEX order so so why shouldn’t they get Kaiser deference in this context?
Wall (Citadel Lawyer) 9:58
So I say two things. Judge Rao. First is you don’t get our if the language is unambiguous, and there’s only I mean, the natural meaning of immediately and automatically is exactly what you think. No intentional delay, and no intervention. And that’s exactly what the contemporary news guide said in 2005. If you look at page 534, of the adopting release, it is almost word for word what I just said, you are right, they came back in 2016. They read in a de minimis exception that has no tie whatsoever to the regulatory text. So first, I don’t think the text is ambiguous enough to allow them to do it and to Justice Kagan already told us in Kaiser that an agency will rarely get deference if it flipped, as it did here. And boy, this doesn’t seem like a very good candidate for that at all. Because when they flipped in 2016, all they talk about is the purposes of the act. They don’t make any effort to tie it to the text. Even the discussion of policy is pretty thin. And then when they revisit it here in 2020, Judge Rao and this is at the top of page 83 of the Joint Appendix. It’s literally two senators. And only the second Senate does any work. And all it says is we looked at this in 2016. And for the same reasons we’re doing it here, except that here they’re doing it for display prices. This is supposed to be the prices that the market tells you the fair value is and they never looked at with a repricing those frustrates the Exchange Act and its purposes, would sit
Would Citadel, though, didn’t challenge the interpretation in 2016. Did it? Well,
in all fairness, not to this court judge Rao we did challenge at the Commission, we put in comment letters pointing to the regulatory text now, but at the time, they said, We’re only doing this with non displayed, right, the stuff you can’t see that you just want to go fishing for in the dark. But we’re not doing it to display prices. So we didn’t petition this court for review. It wasn’t until they did that here that we’ve come in, obviously, and brought it to this court. And at the end of the day judge Rao , if I’m right, that they can’t take advantage of power, then it’s just the the regulation. And what does it mean for a quotation to be immediately and automatically executable, it means you can’t slow it down at the exchange. And once it hits the exchange, you can’t tinker with it. That’s the whole meaning of immediately and automatically. And what they’ve done is say well, not immediate, but as long as it doesn’t take too long that we think it frustrates versus the act, and they flipped the default have automatically never interpreted that word. Instead of automatic execution. You get automatic repricing. It’s the exact opposite of what the regulation requires. So you look, we have both the frontline argument by the way that it’s not consistent with REG but but the fallback argument by the way, Judge Rao doesn’t attack at all 2016 Isn’t inconsistent with it, even if this de minimis gloss can somehow be put on the regulation, they still didn’t hear offer any reason basis for saying here’s why extending that to display prices doesn’t frustrate the purposes of the Exchange Act. There’s no analysis of that in the order. As I say it’s two sentences at the top of page 83. And all it does is point back to 2016. Which of course never look at this extension for display prices. That argument we have regardless of you know, what one thinks about 2016
Judge Rao 13:14
Why would be the interpretation of immediately and automatically be different for displayed versus not displayed?
Wall (Citadel Lawyer) 13:22
Well, I don’t think it was our frontline argument, or frontline argument is they got it wrong in 2016. Fine, they get the benefit of that for the purpose of the exchange. But if they’re going to extend it, that’s still the basis for their order. And this court is required by the APA to say you put a gloss on this that is just not consistent with the regulation. But on the fallback argument, Judge Rao? Absolutely. You could easily say, Well, look, even if you think repricing doesn’t affect for the stuff over here in the dark that you can’t see, it does very much frustrate the purpose of the Exchange Act to do it with respect to displayed liquidity, right, because the whole idea of this protected quote system which doesn’t apply to non display, it’s only for displayed quotes. The whole idea is you got a best bid and a best offer. And those are driven by market forces. And once you have them, everybody’s required to trade at them. You can’t mess up the investor by trading at some different price than you could get on some other exchange. But the whole system word the reason you offer that insurance to the investor is because you’re confident that the bid and offer are actually the best prices out there in the market driven by market forces. Here they’re requiring us to treat these things like they’re the best bid or offer on the market. When we know that when we’ve got a large order, we’re going to get to iEX in the very last piece, and it’s going to fade away from us, right? We have to treat it like it’s real when we know in fact, it’s Phantom and that’s at odds with the system. And if if the SEC wanted to try to do it for displayed prices, it at least needs to explain why it doesn’t frustrate the purpose. was just the act and it didn’t even try here.
Judge Rao 15:05
My colleagues have what do my colleagues have any questions? I have one more question. So I guess I wanted to understand more about this, this question about asymmetry. So all incoming traffic has to cross over iEX is speed bump? That’s correct. Right. So it’s symmetrical as to the speed bump. Um,
Wall (Citadel Lawyer) 15:32
well, surge probably was for the non displayed, right. But this automatic repricing makes it asymmetrical, because, you know, you put the quotation up there. And while everybody else is trying to hit it, and get through the speed, bump, and execute against it, they’re automatically repricing during that 350 microsecond delay. So it’s now asymmetric. And it hurts just the liquidity taker side of the market.
Judge Rao 15:56
But doesn’t I mean, doesn’t the SEC routinely approved? effects? I mean, that I mean, isn’t that the same difference here as it would be with other types of, of orders that move in that way?
Wall (Citadel Lawyer) 16:13
So I don’t think so. Judge? Well, I’m not here to tell you obviously, the question is whether it’s unfair discrimination under the act? Sure. There are some kinds of discrimination that are not unfair. But it seems to me if the commission wanted to justify this, and we know it’s discriminatory. W e know it helps liquidity providers and hurts liquidity takers. I don’t think anybody doubts that. So the question is just is it unfair? And it seems to me, they’ve got to do two things. They’ve got to be right, or at least fair and reasoned about the benefits, right? What they’re attacking, they’ve got to correctly understand the problem and what they get from it. And they’ve got to correctly understand the costs. And here they messed up both sides, right, they took as evidence of latency arbitrage something that’s just evidence of normal, large trading. And then they assumed away the cost and sort of said, well, you can deal with it through routing, but not looking at the cost of rerouting which, as I tried to explain earlier, Judge Walker real. So if they done that on both sides of the ledger, maybe as I said to you earlier, they’d be able to get away with a proposal like this. But in no event, can they protect it, they have this innovation story, they want to tell, that’s fine. If they’re right about this, let the market decide liquidity providers should flock to iEX if they’re right. But what they can’t do is let the platform itself start monkeying around and interfering with the trades, and then claim that those are protected not without a change to their rule, their rule says immediately and automatically, the whole idea is, if you’ve got the best bid or offer driven by the market, you get protection. It’s not driven by the market here. It’s driven by a formula that the market is using to readjust the price that is the result of market forces. So if they want to do this, let them do it. But let them compete as they say they want to don’t let them do it, having interfered with the market and then claim this protection as benefit of rule 611.
Judge Rao 18:07
And so your argument is that rule has to be updated through notice and comment rulemaking and SEC wants to update it.
Wall (Citadel Lawyer) 18:15
Yeah, I mean, look, if they want to say it doesn’t have to be immediately and automatically executable, you can have an intentional delay, and you can have intervention by the agency. It’s perfectly fine. They’ve just got to choose different words. But again, Judge Rao, you don’t have to agree with me on that. Because even if even if I’m wrong, and there is some de minimis exception that can be read into that language, notwithstanding its natural meaning, they still here did not do what they tried to do back in 2016. They didn’t say, Okay, we’re extending it to display prices. Here’s why that extension doesn’t frustrate the purposes of the act. They pointed back to 2016. But of course, in 2016, they were never looking at displayed prices. So they’ve got an APA problem anywhere they go. Although I will say I think that it’s more fundamental than they recognise.
Judge Rao 19:03
My colleagues have any other further questions? Okay. Thank you. We will. We will hear next from the SEC.
Emily True Parise (SEC) 19:14
Good morning, and may it please the Court Emily true Parise for the Securities and Exchange Commission. The Commission’s order in this case, approving iEX, his proposal to offer its customers a new order type was the product of reason decision making supported by substantial evidence and should be affirmed. I would like to go straight to this idea that it’s large orders that are turning on the the CQI which is something that I heard from petitioners Council, the committee, Lord orders that were turning on the CQI. What the commission found looking at the data in the record is that it looks specifically at the extensive data in the record for what was happening at the time that CQI in the small milliseconds and microsecond And it found this that stark disparity between the volume of marketable and displayed orders that were being executed versus non displayed. And it’s important to unpack that a little bit because the displayed versus non displayed shows that it’s it’s the activity is targeting those the borders that don’t have the protection of the of the speed bump in the CQI. And the marketable orders. So is that it’s the aggressive orders taking it, it’s not the midpoint orders, it’s the ones that are crossing the spread aggressively. So there was extensive data in the record showing that it that the CQI only turns on when its aggressive order taking the hallmarks of latency arbitrage, then you look at the qualitative data in the record that the Commission had before it. Right, numerous come. Numerous commentators wrote in saying that they could not trade in that manner, and that those that do use these large orders, these inner markets with orders would not have any problem filling, because the commission made extensive findings that the in the accounting for this de minimis speed bump is a speed bump is something that is commonplace in order routing, there would be no additional cost, because it’s commonplace order routing, that people already have to take advantage of, but take account of because the the just the delay, it iEX is no different than the geographical latencies they’re in here and all the different exchanges,
Judge Sentell 21:25
do you what do you what do you say to cancel arguments that this is inconsistent with the immediate and automatic language that of the existing regime?
Emily True Parise (SEC) 21:36
Sure, I’d be happy to go to that argument, Your Honour. So we think the language really here is right, clear. And I think it’s just a misunderstanding of the of what that symbol 611 inq uiry is getting at here. Right? So rule 611, is, says that, you know, protected, quotations can’t be traded through, and then you have to use the definitions to figure out what protected me. And so in order to be protected, it has to be automated. And that’s really the question here is are all are these quotes, automated, it’s important to take a step back and think about this, there’s only two choices, it’s either automated or manual, right? So and it simply makes no sense to think of these quotes as manual. That’s your thinking, you think of a trading floor, and a human sort of intervening. These codes are not manual, they’re automated, because they immediately and automatically execute. And to focus on the word immediately, as I heard some of the discussion focusing on it cannot be that it has to be immediately has to mean, literally, there’s no delay at all, because all quotes experience some delay, and the commission would explain this very clearly. And it’s automated quotations interpretation issued in 2016. Right, all quotes have some geographic geographically, and so you can’t take immediate to mean, it’s literally mean, there’s no geographic delay whatsoever. So once you’ve taken account of the premise that there’s some geographic delay, the question is, well, is this the minimis? Is the delay any different? It’s not? It’s the commission found it’s well within the same kinds of geographical latency limitations that all other protected quotes get. So it’s, it’s quite simply consistent in this case.
Judge Rao 23:09
My my question about that. I mean, immediately, I mean, isn’t the natural meaning of immediate? I mean, sure, there might be a geographic latency. But if someone takes an intentional action to impose a delay, even if that delay is very small as it is here, I mean, doesn’t that mean that it’s not immediate? So when there’s a geographic latency, you know, immediate is just sort of limited by technology, you know, how fast how immediate can something be, but if someone takes intentional action, and intentional delay, that doesn’t strike me as consistent with the natural meaning of immediate
Emily True Parise (SEC) 23:47
so your honour Citadel made that exact argument in 2016, in their comment letter about the plain meaning and the Commission addressed it in 2016. And it said, There’s nothing about the intentionality of the delay, because it’s simply the same from the perspective of the execution as the geographic delay. And so to the extent that the intentionality piece comes in, it does, it just doesn’t come in in the in the rule 611 out analysis, because the rule 611 is simply about the duration of time, and the intentionality piece doesn’t come in there. It may come in intro to
Judge Rao 24:18
the protected quotation side of things, right? Because Because the idea is that it’s a protected quotation, because because of market forces, not because of, you know, not because of an algorithm, as Mr. Wall said earlier.
Emily True Parise (SEC) 24:34
Well, it’s protected because it’s, I mean, it’s it’s sort of just ipso facto, from the definition is protected, because it’s automated. And it’s, it’s automated, because it immediately and automatically executes and it’s important to go back I think, your honour to answer your question to the to the reg NMS and what 611 was supposed to do. The idea there was to encourage automated trading as opposed to manual trading, right. That was the specific purpose of the trade through exception, in 2005, gave the protected quote status to automated trading as opposed to manual because annual was too slow. And so the idea was to encourage automated trading, and there’s just nothing that’s not automated. I mean, it makes no sense to think of iEX as trades that I expose, as not is not automated. It also doesn’t, it just doesn’t make any sense to think of only the D limit quotes as not automated and therefore protected, because then you have a strange situation where some of iEX quotes are protected, but only the D limit ones aren’t. And the people trading don’t know which is which. So it just, it just doesn’t fit under the structure of the regulation itself, to not consider these these quotes protected, and therefore automated the questions about access to the liquidity and and what Mr. Wall was referring to as phantom liberty, that is addressed extensively by the Commission, but under the Exchange Act analysis, as opposed to the rule 11. And that’s usually the rule 611 analysis, and the Commission found that there would not be extensive court fading and everything that that that goes to that analysis, but the rule 611 analysis is really very clear. Under the terms of the statute, you’re either you’re either automated or manual, you’re automated. If you’re automated, you’re protected under Rule 611.
Judge Walker 26:18
What do you say to the the argument that the SEC itself caused the problem of latency arbitrage that iEX is going to elaborate links to saw, because the cause of latency arbitrage was reg NMS.
Emily True Parise (SEC) 26:42
Your Honour, I think I would say that, you know, the issue before the commission here is whether the order is consistent with the Exchange Act. And so the commission has to take these things as they come. And you know, the commission is bouncing competing policy concerns in terms of the Exchange Act, and the 11 Cafe and the congressional directives that are set up there to to create this national market system. But the issue here in front of in front of the court is simply whether iEX this proposal to offer this New Order Type to its customers is consistent with the Exchange Act. And here the Commission found that it clearly was there was extensive data in the record, and I didn’t want to touch on I see my time. My time is up. And so I just wanted to quickly touch on if with the court’s permission, the idea that the Commission ignored it and doubts data, it simply did not aged for you do
Judge Walker 27:33
that, though? I understand. It’s it’s not perhaps directly related to the question before us, but it does seem to inform it. Do you agree that reg NMS was sort of the first mover the cause of all this latency arbitrage trouble that Citadel says doesn’t exist, but iEX says does exist and SEC does exist and iEX is trying to fix?
Emily True Parise (SEC) 28:04
Your Honour, I think pinpointing the cause of you know, a market phenomenon like like latency arbitrage is a very difficult thing to do. And I don’t think the commission has said anything of that. I mean, certainly, you know, regulations can have consequences, both intended and unintended. But I don’t think that we necessarily would agree with the petitioners characterization there.
Judge Rao 28:25
Miss Parise, can I use since briefing in this case, have other exchanges sought to follow iEX as model? And if so how is the SEC handled those proposals if there have been any,
Emily True Parise (SEC) 28:39
I don’t believe there have been any, occasionally, someone will copy a certain feature of the plan, but then they sometimes they will take them back. But I don’t believe there is any copycat of this proposal in the market. And precisely because to do so you would need to already have the infrastructure. You’d have to you’d have to speed up speed bump, and the CQI and you know, all the pieces that go up that lead up to the D limit orders and there simply aren’t any other exchanges that have all that infrastructure. So I don’t believe there are any other copycats. Thank you. Okay, ask the court has any questions?
Judge Rao 29:20
Thank you very much. Well, we would ask. Thank you. We’ll now hear from Miss Stetson.
Cate Stetson (IEX Intervener) 29:27
Good morning, Your Honours, and may it please the Court I’m Cate Stetson representing iEX as the intervener. I want to make just a couple of points on the primary argument, in this case, the latency arbitrage issue in the solution that IEX devised. And then I do want to be sure to talk about this immediacy issue. Judge Sentell and judge Rao that you raised as well. But just on the latency arbitrage point to begin to judge Rao. I think one of your questions pointed at this as well. This primary argument of citadels is a substantial evidence argument. That’s unusual. You Usually there’s a lack of statutory authority argument or some other more significant APA hooks. So here what we are looking for is, was there substantial evidence in the record that latency arbitrage occurred? And the commission found? Yes, those sites include Joint Appendix 6287 89. When was latency arbitrage happening? The Commission concluded based on IEX’s several months worth of data, that it is happening in those microseconds, right before a price changes. And I want to mention something that Mr. Wall said about the dogs barking if there is data of latency arbitrage. There are so many dogs, there are so many comments, in this case, from hundreds of investment advisors, and companies representing trillions of dollars under management and millions of investors, including, among others, high frequency traders, xtx and virtu, virtu in particular does precisely what Citadel does, we can talk about, you know, what we’re calling the retail issue in a minute, if you all like but understand that virtu supported this D limit order, and does exactly what Citadel does. With respect to Mr. Wells other comment about this evidence that iex put in over several months of data in which the commission evaluated and accepted as evidence of quote, normal, large trading, that is simply not true. What we are talking about is evidence of a huge amount of trading going on in the couple few microseconds before a price changes. I just happen to blink my eyes, that Blink is about 200 times longer than the couple microseconds that we’re talking about. So when Citadel talks about routing, 15% of its batched retail trades at that very moment, that is Citadel trying to hit that stale quote, that is what is happening. It is not evidence of normal, large trading. With respect to the data that Citadel put forward. And I want to make clear, there is a difference between disagreeing with a commenter and ignoring a commenter. And I think a lot of what citadels brief tries to do is to is to shoehorn this into some model where the commission didn’t ignored citadels data didn’t pay sufficient attention. And there is a huge amount of work that the Commission did in this order to take into account that data and to talk about it and explain it. And one of the things that was briefly mentioned, I think, is important is that data comparing the difference between what’s happening on the displayed order side, and what’s happening on non displayed the other thing that I would mention is the Commission’s finding, and appendix 55 to 56. That the people who are doing those trades, the entities doing those trades in those microseconds, are proprietary trading firms, high frequency traders, those high frequency traders include Citadel, Citadel is trading on behalf of retail investors. That’s the phrase it uses Citadel pays millions, hundreds of millions of dollars to brokers in order to get those retail orders so that it can batch them, route them as it sees fit, internalise them if necessary, and profit off of them. That’s citadels model. And that’s why the retail investor was really just a red herring. Let me stop briefly.
Judge Walker 33:46
Yes. Just to make sure I understood what you just said. Are you saying that when Citadel says that 40% of its trades on iEX or on behalf of retail customers, that in fact, those trades could be using latency
Cate Stetson 34:01
arbitrage tactics? 100% true.
Judge Walker 34:05
Yes. And then I while we’ll we’ll address that in rebuttal,
Cate Stetson 34:08
if he disagrees. That is exactly right. And that explains, you know, as I mentioned, Judge Walker, why such a vast percentage of those trades is happening in those couple microseconds. You have dozens of commenters saying we cannot use the technology that these high frequency traders have in order to do this. There’s a comment from famous trading that says we are in the middle of a speed war that we never signed up to fight. The only people who are able to exploit those microseconds are high frequency traders, like Citadel and like virtu which supported this this order. I know my time is up, but I do want to briefly,
Judge Rao 34:47
I do have one question Citadel suggests that really, it’s iEX that’s engaging in latency arbitrage by repricing you know by Reaper automatically repricing when the indicator is triggered, and what do you make that that argument.
Cate Stetson 35:01
You know, I think to begin with, I think it’s a little bit rich because I think as even some of citadels and Miki acknowledge, and I’m talking about the vomer brief at footnote five, there have been a number of innovations that other exchanges have implemented, including NASDAQ nicey, that are designed to either supply colocation, more auspicious location for a fee, or faster data connections for a fee. So far from latency arbitrage, what iEX was founded to do was to actually try to level the playing field so that the daily investors, the people who are actually putting retirement money into retirement funds that are managed by these long term fundamental investors, so that they have a shot, because no one else is able, as I said, to exploit those microseconds except for high frequency traders. The other thing I guess to mention judge Rao is when that price crumbles, that is what iEX is doing in those moments is essentially fighting fire with fire. Because the reason that the high frequency traders are able to exploit those microseconds those tiny moments are because they are anticipating the price change. That’s exactly what iEX is doing. So to say that we’re engaging in latency arbitrage, when we’re actually trying to level the playing field, I think is a bit of a misdirection on the immediate and automatic point. I don’t think Judge Rao that it’s right to say that a speed bump delay like iEX is somehow qualitatively different than say a geographic delay. Exchanges make choices all the time about where to locate their servers should be located in Mahwah, Chicago, Miami, exchanges make choices all the time. And they build their high frequency traders all the time for even locating servers in different parts of a room in order to make sure that those servers are more quickly connected. exchanges for ages have tried to make sure that to the extent everybody’s paying the same price, but are located in different parts of the room, that there’s extra cable, in order to make sure that there’s no tiny latency between the server at the north end of the room and the server at the south end. All of these are examples of you know, what we would call geographic latencies. But they’re just as intentional as anything that iEX designed. And what the commission concluded in 2016 Was that the immediate language means so de minimis, that it doesn’t interfere with fair and efficient trading. Now, the argument Mr. Wall offered, Mr. Easy is exactly right. This is exactly the argument that was offered in 2016. That immediate means immediate, immediate, though, does not mean instantaneous. If you, Judge Rao asked me to come to your chambers immediately, it would take me a while to get there. If you asked me to come to your chambers instantaneously, I would say I’m sorry, I’m not capable of doing that. There’s a difference between immediate and instantaneous and that difference is within the Commission’s bailiwick and prove you to interpret, you know, what it came down to in 2016 is exactly what it comes down to today, which is does this de minimis delay, interfere with the fair and efficient execution of an order? What iEX is,
Judge Rao 38:33
if your view is that’s in that immediate is ambiguous, and so we should defer to the SEC’s interpretation?
Cate Stetson 38:41
I think that is the SEC position. And I can see the the merit in that position for the precise and the reason I just said, I think when people think about immediate and you’re not thinking in terms of these eye blink moments, immediate might mean something very different. But when you’re talking about these eye blink moments, immediate has to mean something that’s a little bit more accommodating. It’s exactly why the commission even with its protected quotes, builds in a one second, essentially grace period, no one is is found to be in violation of the rule as long as that quote changed a second or less before, you know, whatever order they executed. There’s an acknowledgment that the systems can only go as fast as the systems can go and hear where you have targeted I’m sorry, Judge Santo
Judge Sentell 39:34
It just bothers me that you just if what you just said your space is the system’s can go. The geographic delay, if you would between immediate and instantaneous, recognise is that effective system can go here you are deliberately putting a garden you’re not the the rulemaking is deliberately putting a delay into the – course with a transaction that takes it out of the instantaneous, but may also take it out of the immediate. Isn’t there a difference between the geographic delay that is inherent and unavoidable and the inserted delay of the roadblock, which is evitable?
Cate Stetson 40:19
Judge Sentell two answers. The first is that as Miss Parise mentioned, that is exactly the argument that was made in 2016, in which thoroughly evaluated and rejected the second answer is, as I just mentioned, to judge Rao, the the idea that geographic latencies are unavoidable is simply not the case. Exchanges make choices all the time about where to locate their servers, how fast to connect their particular customers to a particular server were in the room, as I said, to locate the server, those are all geographic latencies of some degree or another. So drawing a distinction between the kind of coil delay that we’re talking about here with iEX. And the kind of coil delays that are common in other servers, including, as I said, to try to equalise high frequency traders access, if they’re are slightly different from each other in a room such that one gets to another, you know, one microsecond before, all of those are intentional choices. And what the commission concluded in 2016, and reiterated here is that word immediate in this context means a delay that is so de minimis. It’s not sort of human. There’s no human interference, which was the whole reason why this immediate order began to exist. It’s so de minimis as not to interfere with fair and efficient access. And I think what Mr. Walls client is seeking, is what we would say is unfair access, because they are seeking to exploit that tiny moment of time when a price is changing before other people, other liquidity providers and liquidity takers are able to get there. And I want to perhaps end on this if there are no further questions. This is discrimination against a type of high frequency trader that engages in predatory latency arbitrage. That is the discrimination if we want to use that word that we’re talking about what the SEC concluded what iEX submitted, and what the SEC evaluated and adopted was that this D limit order helps the market overall, it helps liquidity providers, it helps market makers. That’s why you have XGX and virtu joining these comments. It helps liquidity takers because it ensures that liquidity actually exists. And isn’t just shifting over to the dark markets or off exchange. This order helps the market overall, the people it doesn’t help are the people who are trying to essentially skim that predatory tax off of a changing price. If there are no further questions,
Judge Rao 42:58
any further questions? Okay. Just sorry. Yes.
Judge Walker 43:04
You, were you saying the session that immediate immediately is ambiguous, or you win if it’s ambiguous, and you also win if its unambiguous?
Cate Stetson 43:14
The latter. It I think what what the SEC said in its brief, I think in a footnote is that the the interpretation is owed deference under Kaiser. That’s clearly true, just given the length and the thoroughness of the consideration. The the specificity of what we’re talking about the technicalities of what we’re talking about. But if it’s unambiguous, then we win also. So the answer is both.
Judge Rao 43:44
Thank you, Ms Stetson. Thank you. Mr. I’m not sure you have time left, but we will give you an additional two minutes for rebuttal.
Wall (Citadel Lawyer) 43:53
So, three quick points. Let me start with rule 611. You’re right, Judge Walker, regulation and MS did create the fragmentation the protected quote, stitches it back together. And it’s important to enforce it. The language of the regulation is not anything about interference with fair and efficient trading, it says immediately and automatically executes an order and your right Judge Rao. Immediate, of course, things take time to do and it doesn’t mean you’re not doing them immediately. But when you introduce an intentional delay, it’s no longer immediate. And when you tinker with the quote, it’s no longer automatic. And in case
Judge Rao 44:24
you respond to miss Stetsons point that it’s actually that there are intentional ways that that traders affect God Graphic latency. So it’s not really qualitatively different to have something like a speed bump.
Wall (Citadel Lawyer) 44:40
Sure, so two points. One, it’s not true in practice, as we explained to page 372 of the JA, when we have a large order set, it all sends it out to the exchanges simultaneously. Sure, there’s a little bit of geographic latency as you go to the different exchanges. But the difference isn’t differences and large enough to allow anybody else to see what’s happening on the other exchanges, and react. So because we have a best execution obligation, we send it out everywhere at once. This is lengthening the delay, as Judge Sentell said, and it allows everybody at iEX to see what’s going on. And then they reprice. And the second thing is, it’s just not a natural way to use a language around. Sure. I tell my kids to immediately make their bed in the morning, they got to get out of bed turn on the light. Nobody thinks they didn’t comply with my instruction. But we sure do if they get up and read books or play with toys for a while. And in case you think there’s any doubt about that, all you need to do is look at the adopting release with rule 611. It says at page 534. It’s not automated. If you have any other type of intentional device that would delay the action, the Commission read it exactly the way you do.
Judge Walker 45:41
And it seems that Citadel pays for those devices on on several other exchanges. So tell me tell me if I have the facts wrong, I’m imagining Citadel and Company X that are equidistant from an exchange. And so they send they send a communication to the exchange at the exact same moment. But Citadel has paid extra money to that exchange, so that its receiver is a little closer than the competitors receivers, I’m imagining a room with computers in it and there’s like a box for Citadel and a box for the competitor insurance is probably vastly oversimplified. But Citadel pays the exchange money so that its receiver is in a better position, a better distance, a better location in that room, then Company X. Does that happen.
Wall (Citadel Lawyer) 46:40
So John Walker, I think maybe it did years and years ago. And it’s a wonderful narrative. I grant you look, it’s a great story, the myths that’s untold. But no, in this day and age is not true.
Judge Walker 46:49
First of all, and I you know, just it’s not. We had a case last month about sky beams, communicating between mawah and the other things. I don’t think there was any party in the case, they disagreed about a lot of stuff. But none of them disagreed that scenario and just happens where are companies pay a little bit more money to an exchange like NASDAQ or New York Stock Exchange to have their little box closer than everyone else’s little box? It’s very expensive real estate?
Wall (Citadel Lawyer) 47:19
Yeah, no, no, Judge Walker, what I was gonna say was some exchanges deal with this by having standard length cables, and the exchanges that don’t have that everybody pays to co locate. So XTX, virtue, two sigma, all the firms that Miss Stetson that are citadels competitors. They’re all right there, there is no latency. I’m not saying they’re not paying for it. I’m saying there’s no competitive advantage that we give rise to latency arbitrage. Because everybody’s got the same connectivity, and everybody’s got the same data feed, they may be paid
Judge Walker 47:50
off. There’s no colocation at an exchange that’s slightly better than another colocation at an exchange.
Wall (Citadel Lawyer) 47:56
Judge Walker, I don’t want to go that far. What I want to say is my understanding is that’s not a common problem across the exchange when you talk to traders. And at a minimum, the Commission didn’t put on that kind of evidence. That’s the interesting thing about the order. If that’s true, what the what you’re playing out and what Ms. Stetson said, it ought to be easy, right? We ought to come in with examples. But there’s facts on the ground, right? We could just show where the servers are, and all the colocation stuff. The Commission didn’t do that. And it didn’t even try to take actual trades, and say, here’s the latency arbitrage. It looked at aggregate statistics and said, See, those show that there’s a lot of stuff going on when the signal is is turning on. And this was the second point I was going to make in the row, which is, if you none of that shows, what’s turning the signal on. So Miss Stetson is right that a page 88 of the Joint Appendix they say, large orders don’t turn it on. But when they say as shown by the data, above all that to pointing to on pages 87 and 88 is the stuff about there being a lot of trading when the signal was on. And we put in this actual order at page 372 of the Joint Appendix, we said here’s an actual order on behalf of retail investors. It triggered the signal just because it’s a large order has nothing to do with arbitrage. And as I stand here today, having read everything in this case, I still don’t understand what the Commission’s basic responses to the fact that this is the kind of trade that turns it on, because it makes sense, right? You’re you’re taking a lot across a lot of exchanges. iEX is watching that, of course, it’s going to predict that the price is going to move the signal is going to come on. By definition. It’s probably a good size trade or good number of orders, right because you need a whole lot of liquidity This is a small player doesn’t have a lot of liquidity. So then it says, Oh, we have a lot of things going on in our shop when the signal comes on? Of course you do. It’s exactly what you would expect all of the mark out data, the spread crossing, it’s exactly what you would expect to see if large orders turn the signal on. And so I would just point back to the Commission’s order. And the third thing I would say, Judge Walker to go to your colloquy with with Miss Stetson. That’s not where the SEC started. Miss Parise, said, look, it’s commonplace to do this as a matter of routing. That’s not true Citadel sends these orders on behalf of retail investors to exchanges simultaneously when it’s a large order. We explain that to commission there’s nothing to the contrary, in the record, could we reroute and hold back orders and go to iEX? First, we could with the costs I explained earlier, either higher costs or missed executions to retail investors who won’t get the bulk of what they want, because the iEX tail is wagging the New York Stock Exchange and NASDAQ dog. Now Ms. Stetson takes it another direction. She says, Oh, well, they’re batching. They’re bundling. And so it’s evidence of latency arbitrage. And with all respect to miss Stetson, that’s false. We told the commission it’s false. I’m happy to supplementally brief it, we have best execution obligations, Citadel does not batch and try to time retail orders. It may feel some of them internally.
Judge Rao 50:49
Mr. Watson is that how does that work? If 15% of the retail orders are coming in when the signal is on? I mean, that how to how do you explain that?
Wall (Citadel Lawyer) 50:59
Oh, so that that part, I think is fairly straightforward to address it. It also handles more retail than anybody in the country, we have 40% of the retail orders on behalf of investors out there in the country, right. So we have a lot of retail, and we have some small orders and some one offs. And sometimes iEX is good for that. It doesn’t have a lot. But sometimes it has a little bit of kind of random things. So you send a lot of small orders there. But the large orders that we sent, that’s the 15%. And what we found is that, that 15%, were turning the signal on, as we explained 99.3% of in volume were large orders. Once our retail orders got large enough, they triggered the signal, and it came on. And the Commission’s response to that at page 88 is just Ypsi Dixit, all it says is Oh, but look at our data. There’s a lot going on when the signal comes on. Of course, there’s a lot going on, we’re sending large retail orders. That’s the 15%. Others are sending more large orders and fewer small orders than citadel. So that’s why even though it’s only 15%, for us, it creeps up to 24% of volume 33% of orders. But it’s all the same phenomenon. And just to close Judge Walker with where I was before, these retail orders, they’re not like there’s nothing misleading about saying on behalf of retail investors, the example we gave at page 372 of the Joint Appendix is, is real. Maybe it’s TD Ameritrade, maybe at Schwab, maybe it’s a pension fund, an institutional investor, they want a lot of shares, we don’t hold that back and wait to take it to the market in some moment of arbitrage, we’d be violating best execution obligations. If we did,
Judge Rao 52:34
how did the 15% show up in that fraction, a tiny fraction of a second
Wall (Citadel Lawyer) 52:41
Oh, Judge route, that’s that is the key to the whole thing, right? Right there. They’re not showing up when the signal is on. That’s that’s what the other side, we are triggering the signal triggering the signal because iEX is sitting there and looking at these orders, hit the other exchanges, and say, wow, there’s a spike in demand or a spike in supply. So the offer is going to go up or the bid is going to go down. And so they’re turning on the signal. And then the order gets they’re having hit the speed bump, and they treat that order once it hits as the volume or the order that they’re talking about hitting when the signal is on. So of course there’s a lot hitting when the signal is on. They have a formula designed to detect price changes, and they can watch what’s happening on other markets. It’d be sort of like, of course, there’s a high correlation. Of course, you see crossing the spread. Of course, you see a lot of orders coming in, of course you see mark out data showing the prices running away from the liquidator that’s
Judge Rao 53:35
just a race between the technology of your client and the technology of iEX to go back and forth. I mean, isn’t that ultimately in the market? What’s happening?
Wall (Citadel Lawyer) 53:46
I don’t think so. It might be a race between Citadel and two sigma and XTX and virtu and all the other high frequency traders that as I said, Judge Walker have the same kind of activity and same data. But no other exchange is doing this, right. This is pretty novel. Exchanges are meant to be platforms. They’re meant to facilitate trading. This is an exchange, the only one right that’s doing this that’s reaching in and repricing the quotes themselves. And that’s a sort of novel thing. And to get back to where I started. If you conclude on the first issue that they can have the D limit order, even though they haven’t shown you real evidence of latency arbitrage. And even if they have Judge Walker, they haven’t looked at the costs. They’ve not looked at the cost and retail and the costs of routing. Still during the argument. No one has addressed the costs of these rerouting changes but even if you can clued that you can’t have it be protected under Rule 611. For the reason that Judge Sentell got out earlier, it just doesn’t square with the language of the regulation, right protected quote is meant to stitch back that fragmentation, you were talking about Judge Walker, and it does it by looking at the best bid and offer as a result of market forces. If they want to tinker with those, we don’t think they should be allowed to. But what they can’t do then is say, Well, that should be treated the same as every other bid and offer that is actually the result of market forces. Let them compete in the open marketplace. If liquidity providers like it, they’ll go there. And liquidity takers will have to go to get the liquidity. That’s what Canada did when they did the same thing. No protected quotes, let everybody have incentives to use the exchange and then decide let the marketplace decide. And
Judge Walker 55:24
I’ll stop Judge Rao
Judge Rao 55:28
Judge Walker if you have another question, please go ahead.
Judge Walker 55:30
Ask one just one more than the the your last statement about just letting the market for, you know, play out. Here. It seems to be that’s what iEX wants to do. And it’s you who is going to a federal agency and saying stop a private entity iEX from doing what they want to do. You’re the one who’s trying to kind of regulate your way into a market victory.
Wall (Citadel Lawyer) 55:59
So I all respect Judge Walker, I think we see it a little bit differently, right, which is we we don’t want the exchanges that’s right interfere with the market forces themselves. private parties in the marketplace through buying and selling should be determining prices, not the exchanges, the exchange acts as they are meant to facilitate a fair and open market. Not that they’re meant to pick winners and losers and reprice their quotes. But even if you think I’m wrong about that, I think clearly they’re not pro market force on the rule 611 side, if they really believed this innovation story, they would do what CBOE did when it submitted its proposal, it didn’t say it’s quote should be protected. It said, let us do this. The Commission disagreed it flipped here for reasons it hasn’t really addressed. But what iEX wants is the best of both worlds, it wants to be able to do the innovation, but then say that these boats are protected notwithstanding the regulatory text so that it can lock liquidity takers into coming there, which drives up its revenues. Because let’s be really clear, Judge Walker iEX makes money in two ways. It makes money by having protected quotes at the best bid and offer. And by having more trading volume. This change hasn’t driven up in the real world, it’s actual executions. It’s not actually helping the market. But it does drive up the amount of liquidity it has at what’s called the NB Bo the best better offer. And that means more money by the millions for Ms. Stetson’s client, if they want to do that in the marketplace, let them do it. But they should not be able to do it by virtue of rule 611. Either because the regulatory text is clear, or because they can’t qualify for Kaiser because they flipped or because at a minimum even if the de minimis exception exists. They didn’t analyse it here in this order with respect to display prices. At a minimum, this Court should vacate with remand with vacatur of that part of of the order and obviously we have not discussed today the remedy but you know, I think clearly this is the standard if you if you’re going to remain you should vacate Thank you very much. Case is submitted. Thank you
