IEX Group Gains Approval for Stock Exchange | Hacker News Reader

IEX Group Gains Approval for Stock Exchange

nytimes.com

85 pointsanotherhacker36 comments

Can someone tell me if the view in "Flash Boys: Not So Fast" is accurate on IEX? In "Flash Boys" Brad Katsuyama says some ridiculous things, showing he doesn't know how exchanges work. In fact, the entire book is basically complaining that they can't trade large blocks without changing the price.

Is IEX actually doing things in good faith or are they just playing off of people that are scared?

Well to be fair, those were Michael Lewis's second hand telling. I personally know the author of FB:NSF, but haven't spoken to him in a while. I think I agreed with 95% of what he wrote. I think IEX will likely stumble, but will continue. The problem is not that IEX is doing something bad, but they're just not different enough to be meaningful. They have been very good at marketing though and brought up a lot of issues, but definitely raised some that are not accurate.

The unfortunate problem (that's hard to resolve) is those executing on agency will always be at a disadvantage to those executing on principal. For example if someone wants to buy 5000 shares, they can route in a way that risks them buying 8000 to get better prices (by sending extra orders to dark pools, etc), someone executing on agency can never do that and so it might not make sense for them to route to dark pools, but almost all of the routers do that (fidelity, schwab, etc) so they disadvantage some of their customers (although they do advantage the smallest customers). Basically I think ultimately the reg-nms order protection is probably a little too rigid of a structure to fit the diversity of behavior into. I'm not really sure what they're long term hope is. I do find equities to be basically the most fair trading ground on the exchange (which is why they had any foothold at all). It would be nice if they were trying to disrupt the really expensive trading markets like fixed income where the spreads are much wider compared to the risk being exchanged and the dealers have so much leverage that they won't support anybody who is trying to cut into that spread.

FB:NSF? I get the abbreviations, but honestly i'm having a hard time deciphering this one.

Flash Boys: NSF? Please elucidate.

Flash Boys: Not So Fast is a rebuttal to Michael Lewis's book that basically explains why it is wrong, point by point.

https://www.amazon.com/Flash-Boys-Insiders-Perspective-High-...

It's quite good.

Comment below was correct. Sorry, I thought it wouldn't be confusing because it was in comment I replied to. My bad.

It is mostly complete BS. They can't even begin to stop the numerous ways that HFT and sophisticated traders can play with the market, and are putting substantial road blocks that will actually decrease liquidity and increase complexity at the exchange level for basically no reason because of reg NMS integration.

The original plan for IEX sounds like a wonderful extortion racket. By law, you cannot buy or sell stock on someone else's behalf unless you check with all major US stock exchanges first to verify that you are getting your client the best price. IEX intentionally, massively delays answering back unless you pay them buckets of money. Thus, unless you pay IEX buckets of money, you are at a huge disadvantage on every stock market, not just IEX. Quite a racket.

Has anything changed with either the laws or IEX's plan since then?

Update: IEX did indeed change their proposal to remove the favored order routing. Which still affects trades in other exchanges, and still affects negatively effects traders on all markets, but at least its "fair" now.

> you cannot buy or sell stock on someone else's behalf unless you check with all major US stock exchanges first to verify that you are getting your client the best price

It's not just trading on behalf of a client. Even if you're trading directly for yourself, you're still legally obligated not to "trade-through" a displayed price. The idea is to protect the person who placed the displayed order, not yourself or your client.

> IEX intentionally, massively delays answering back unless you pay them buckets of money

You can't pay them to avoid the delay. They delay all outgoing direct-to-trader information and all incoming orders or cancels.

The difference between what IEX does and just placing your stock exchange 38 miles from everyone else's (350us at light speed in fiber) is that while they delay incoming orders and outgoing data, IEX itself listens to data coming in from other exchanges with no delay, and uses that 350us look into the future to re-price orders before anyone else can interact with them.

Yeah, they do plan to delay everyone equally now. However in the original proposal if you used IEX's router, you weren't delayed.

> obligated not to "trade-through" a displayed price.

> The idea is to protect the person who placed the displayed orde

I'm afraid I don't understand what the first quote means and how the person in the second one is protected. An explanation would be much appreciated.

If someone is publically offering to buy a stock for $10.00 on Exchange A (i.e. they've placed displayed limit order), then it's illegal for you to sell that stock to someone else for $9.99 on Exchange B. If that trade did happen, the order on Exchange A is said to have been "traded through". The rule protects limit orders which have the best price available from missing out on a trade just because they were sent to the wrong exchange.

> The idea is to protect the person who placed the displayed order, not yourself or your client.

I'm pretty sure that's wrong. I've never heard NMS/rule 611/etc described in those terms, and logically, it makes no sense. The point of the rule is to try and stop people eg, buying at price X when the NBBO is < X, full stop.

Here's my evidence that Rule 611 is more about protecting resting limit orders than marketable orders:

1. It's called the "Order Protection Rule".

2. It defines "protected quotations" and makes it illegal to trade through them.

3. It doesn't just apply to agency trading (where you could argue that it's meant to protect the client); it's also applicable to principal trades.

4. There's no allowance for trading off fees or latency or probability of getting a fill against price. For example, you can't choose buy immediately at $10.01 on Nasdaq instead of waiting for a 30ms round trip from Chicago at $10.00, because it's the quote on Chicago that's being protected, not you.

5. If you send an ISO order to an exchange (which basically says "don't route this elsewhere even if you see a better quote"), and it turns out that you traded through a quote somewhere, the SEC will fine you, because they're protecting that other quote, not your marketable order.

6. This quote from the SEC: "Many commenters on the proposals ... strongly supported the need for enhanced protection of limit orders against trade-throughs. They emphasized that limit orders are the building blocks of public price discovery and efficient markets. ... by enhancing protection of displayed prices, would encourage greater use of limit orders and contribute to increased market liquidity and depth. The Commission agrees that strengthened protection of displayed limit orders would help reward market participants for displaying their trading interest and thereby promote fairer and more vigorous competition among orders seeking to supply liquidity." [1]

[1] https://www.sec.gov/rules/final/34-51808.pdf

What are those buckets of money?

Also, IEXs delay is less than the delay between existing exchanges because of speed of light limits, so massively seems inappropriate.

You might be surprised (when talking about the U.S equities exchanges at least): http://anova-tech.com/wp-content/uploads/2014/09/market-data.... The LoS distances in NY are on the order of 5-30mi and the lowest latency links are over microwave and laser.

The Chicago Stock Exchange is 15 milliseconds away from all the others. There isn't much trading volume there, though [1].

[1] http://batstrading.com/market_summary/

Derivatives are traded in massive volume on the Cboe which is why the time between New Jersey and Chicago is so important

The CBOE matching engine is in New Jersey, along with the equities exchanges (http://www.automatedtrader.net/news/exchange-news/141860/cbo...). The Chicago-NJ latency race is more about futures traded on CME.

Here are some newer numbers: http://www.mckay-brothers.com/wp-content/uploads/2016/06/201...

CME -> NY4 latency is now slightly higher than that allowed by the speed of light.

> By law, you cannot buy or sell stock on someone else's behalf unless you check with all major US stock exchanges first to verify that you are getting your client the best price.

The side-effects of this law and because of it not being able to achieve a 'fair' price was why IEX was created.

Edit: This is neither a pro nor anti comment. It is a factual comment, of the motivation to create IEX.

I read flashboys too and I don't remember it having anything to do with national best bid/offer laws. Was there something specific?

Opponents of IEX, including the other stock exchanges, have argued that the structure of the new exchange will add unnecessary new complexities into an already complex stock market, and potentially end up hurting small investors.

Yeah they are really looking out for the small investors. How very considerate.

It may seem counter-intuitive, but...yes?

Small investors benefit from HFT and related technologies, which is why, eg, Vanguard is pro-HFT. Investment banks are harmed by it, which is why, eg, Goldman is anti-HFT.

I have a pet theory that the articles about HFT are actually placed [0] by big financial interests. I have personally seen a trader at a bank feed a story to a reporter in a different product. That article was pretty fair, but it was on an uncontroversial topic.

I don't think the goal is to be deceptive, but the biases of the source are going to come through. Traders at banks are going to much more charismatic and gregarious than traders at HFT firms so it's not surprising that their story rings louder.

[0] I'm not talking about explicit quid-pro-quo arrangements, but something along the lines of http://www.paulgraham.com/submarine.html

Maybe. But how about letting the investors and market decide? These exchanges do not want competition and are hiding behind the "protect the small investors" schtick.

> how about letting the investors and market decide?

I basically agree with you, and the market has been deciding; IEX is currently the #2 ATS (colloquially "dark pool", but IEX has been publishing quotes). However, now that IEX is an exchange, their quotes become "protected", and all market participants are legally required to send orders to IEX if they're displaying the best price. It's now illegal to decide that you value a quick response or a higher chance of getting a fill over price and choose not to trade on IEX.

Liquidity in instruments other than equities (options/futures/futures options) is getting worse, not better. Wider spreads and less liquid markets actually do disproportionately affect smaller investors who are doing anything other than straight equity buying and selling. HFT isn't hurting liquidity and isn't nearly the problem it once was, and this doesn't do anything to address the actual problems in dark pools or auctions or the non-transparent part of the market.

The SEC’s decision may not be the end of the fight. Last month, attorneys for Nasdaq argued that the SEC could be sued if it approves IEX, saying the SEC would first have to change its own rules to explicitly allow for a speed bump.

To this, the SEC issued an interesting response: addressing concerns about the legality of speed bumps, the SEC separately said that delays of less than one millisecond are consistent with its Regulation NMS.

According to the SEC [1], IEX's 350 microseconds delay is negligible, and thus the market is automated and the quote is protected. The practical interpretation is that the SEC has set a ceiling for what it deems the speed race among HFT firms (with fiber optics, microwaves, lasers...)

So with this decision, the SEC has capped what technological advancement in trading can achieve going forward, as now a 350ms delay will become the norm, while anything below 1 millisecond is deemed a "de minimis" delay. Which is no good for HFTs, where microseconds can mean all the difference between profit and loss.

[1] https://www.sec.gov/divisions/marketreg/automated-quotations...

The IEX delay is uniform and deterministic, so if you're faster than your competitors by 1 microsecond in sending an order to IEX in response to an external event, you still win. The race that you can't win anymore is with IEX's pegged orders, which re-price without the delay (but they're always non-displayed).

To add to your comment here, IMHO the situation is now worse from a technical perspective on multiple fronts:

* IEX is still FIFO, as you point out about the race in response to external events. If Katsayuma really wanted to wipe out latency arb, he should have introduced some randomization and batching into the matching engine. I'm not sure how that would hold up from a regulatory perspective though.

* IEX quote updates will be 350us slower than any other update because of the additional input latency, which means that their protected quotes will always lag behind the market on a price flip. The amount of time that a price level is still protected on the SIP, yet practically gone, will increase as a result. This will increase the amount of time that DAY ISO orders are necessary for establishing the new price level, and therefore will benefit the HF shops which can leverage DAY ISOs the best. This also possibly creates a moral hazard for noncompliance.

* Other exchange owners have expressed interest in adopting a shoebox-like system, because of the expectation that delayed quotes will _force_ more market participants to trade on that exchange. If multiple systems start generating lagged quotes, the first problem continues to get worse.

* IEX's pegged order type relies on the SIP. But now that IEX itself will become a part of the SIP, the shoebox delay will have a feedback effect of lagging its own peg updates. Likewise, if other updates start delaying their output, IEX pegs may become less useful. I'm not exactly sure what the net effect of this will be (after a few years of other exchanges adapting)... but I don't expect it to be "good" for the system as a whole.

In general, I believe that artificial introduction of latency into the National Market System seems only to serve special interests: advanced technological traders (usually HFT itself), and the exchanges which enact the artificial latency.

Citadel is not happy:

“Today’s decision will test and potentially reverse the gains in fairness, efficiency and transparency that have been made to our markets over the last decade. We must be vigilant to identify unintended consequences, and firm in our commitment to equitable and consistent treatment for all investors.” [1]

[1] http://www.reuters.com/article/us-usa-sec-iex-idUSKCN0Z32NM

Nothing to see here. The only people affected are other exchanges who get more competition

If the goal is to avoid high-frequency trading, why not use a blockchain-like technology?

When writing a comment about the blockchain, it's always useful to mentally replace it with the words "distributed database", and see if still makes sense.

"If the goal is to avoid high-frequency trading, why not use a distributed database"? Maybe because that would have absolutely zero impact whatsoever on high-frequency trading?

(Also, the goal isn't really to avoid high-frequency trading, and it's quite likely IEX will create more opportunities for high-frequency trading, not fewer, but that's another matter.)

What specific aspect of blockchain technology?

It's slowness?

The transparency of the consolidated order book?

I place orders on Counterparty a lot, which offers a decentralized asset exchange that settles to the bitcoin blockchain. It would be nice settle stock trades on something like that, slow, too slow, but it would be a nice option.

Not the simplest way to solve the problem. May have many other effects.

Reply on news.ycombinator.com