Help me understand Michael Lewis' "Flash Boys"

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
ftobin
Posts: 1071
Joined: Fri Mar 20, 2009 3:28 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by ftobin »

manwithnoname wrote:Jack Bogle has no problem with HFT.
People from the institutional world have only very, very slowly come to know what occurs in market microstructure.
manwithnoname
Posts: 1584
Joined: Mon Jul 22, 2013 7:52 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by manwithnoname »

ftobin wrote:
manwithnoname wrote:the SEC is slow, so what? the rules are still valid.
Remember, the rules are not specific. That's what FINRA and compliance departments add to. The SEC provides general rules, like best-ex. It's not a specific rule. But you have to be abiding by it -- your compliance department is supposed to enforce this.
How does anyone know whether an order is placed without intent to execute? Can't investors change their minds after placing an order?
Certain behaviors are obviously market-manipulating. Huge imbalances flipping a millisecond before the lock-in time is not someone changing their minds. It's naive in the extreme to suggest people don't post orders without intent to execute.
Are the HFT traders abiding by FINRA's rules? I don't see FINRA complaining about non compliance with NMS?
FINRA is like the SEC, just covering a more specific market segment.
And how is HFT hurting small investors who now have access to more accurate prices because trading spreads have been narrowed which was the intent of Reg NMS.
That was not the purpose of RegNMS. Spreads had narrowed well before that.
It seems that none of the regulatory agencies have a problem with the practices of HFT.
You'd be amazed to know how little the SEC knows of trading micro-structure. Buy-side firms almost as as much so.
So? There is nothing of substance to discuss. SEC and FINRA are aware of what is going on. Until they take action their inactivity indicates that they have no objections to current trading practice under Reg NMS. No brainer.
Last edited by manwithnoname on Fri Apr 04, 2014 4:31 pm, edited 1 time in total.
ftobin
Posts: 1071
Joined: Fri Mar 20, 2009 3:28 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by ftobin »

nisiprius wrote:If the SIP doesn't have the actual best price, how do you know whether Fidelity got it for you?
Fidelity is saying they'll provide NBBO price that they know from the SIP when your order enters. It might be a millisecond or too slow, but that's fine. Your order just getting to Fidelity probably took hundreds of milliseconds.
manwithnoname
Posts: 1584
Joined: Mon Jul 22, 2013 7:52 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by manwithnoname »

ftobin wrote:
manwithnoname wrote:Jack Bogle has no problem with HFT.
People from the institutional world have only very, very slowly come to know what occurs in market microstructure.
Please explain the take away from your post because it is not obvious. Are you saying that Jack Bogle doesn't know how HFT is affecting VG and its customers but you do?
ftobin
Posts: 1071
Joined: Fri Mar 20, 2009 3:28 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by ftobin »

nisiprius wrote:So: 1) if using the SIP is optional, what was supposed to be the point of regulation NMS? 2) Is Hunsader correct that what the exchanges are doing is not just unfair, but "actually illegal under Reg NMS?"
The point of RegNMS was to ensure that you were always executed within the NBBO. Exchanges use data from the to ensure they're not executing outside the NBBO. They are making a reasonable effort to know the NBBO, and using an industry standard.

There is no problem that exchanges use the SIP. It's perfectly reasonable to expect there to be a consolidated data source for stocks. We just have to be aware people can take advantage of its weaknesses, and understand whether not other safeguards are needed.
ftobin
Posts: 1071
Joined: Fri Mar 20, 2009 3:28 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by ftobin »

nisiprius wrote:Do high-frequency traders get to cancel their orders for free?
Everyone does.
manwithnoname
Posts: 1584
Joined: Mon Jul 22, 2013 7:52 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by manwithnoname »

ftobin wrote:
nisiprius wrote:Do high-frequency traders get to cancel their orders for free?
Everyone does.
I do. No brainer
User avatar
JamesSFO
Posts: 3404
Joined: Thu Apr 26, 2012 10:16 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by JamesSFO »

Put Michael Lewis' book aside on the details.

What drove the RBC (Royal Bank of Canada) to create Thor to avoid front running from HFTs? And why did the underlying team go create IEX (investors exchange) to prevent the same?
gw
Posts: 776
Joined: Thu May 28, 2009 10:12 am

Re: Help me understand Michael Lewis' "Flash Boys"

Post by gw »

ftobin wrote:
Was the whole magic of THOR that they equalized time of arrival?
As far as I can tell, Thor offers two value-adds. They could act as your router, and hit all the other exchanges simultaneously (they patented this). Additionally, they slow down incoming orders until they have the correct NBBO. This is so that if the NBBO offer was just lifted (increased by a penny), and an institutional order was resting pegged to the offer at IEX, HFT can't come in and execute against the institutional order before IEX realizes the NBBO has changed.
Minor point: I think you're conflating RBC and IEX. Thor was developed by Katsuyama, the star of the Lewis book, and his crew while he was at RBC, and it was patented (somehow successfully) by RBC. Katsuyama later left RBC to create IEX. I don't know that IEX has rights to use Thor (maybe you do?), though it wouldn't be surprising.
ftobin
Posts: 1071
Joined: Fri Mar 20, 2009 3:28 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by ftobin »

JamesSFO wrote:What drove the RBC (Royal Bank of Canada) to create Thor to avoid front running from HFTs? And why did the underlying team go create IEX (investors exchange) to prevent the same?
Well, everyone is trying to create systems to get themselves the best execution. Thor is such an implementation, and the markets are full of competition. Whether or not HFT is doing something against regulation is a separate question.
gw
Posts: 776
Joined: Thu May 28, 2009 10:12 am

Re: Help me understand Michael Lewis' "Flash Boys"

Post by gw »

IlliniSigEp wrote:I really am not trying to be Cpt. HFT, but I guess I will be for sake of discussion!

Jack's house analogy has some issues.

The situation would be more like this, there are 2 houses for sale, both for $100. You walk over to one house and buy it for $100. Your neighbor sees you do that, figures you might be interested in the other one, and runs out and buys it for $100. Now you have a choice, pay your neighbor more, or don't. Maybe you should look in another neighborhood for a $100 house! Or maybe, before you buy the first house, you get a buddy and walk to both house at the same time and buy both. No problem.

An order inflight to a given exchange can't be both seen and traded infront of electronically. The issue is when a broker does a bad job with a huge order and stupidly buys 10,000 shares on NYSE and then tries to by 10,000 more on Nasdaq. Well guess what broker, you have shown your hand to the market. The market now guesses you want more shares and adjusts. If you don't buy, the market is now overpriced, someone will come in and sell it back into place. The issue is really that the buy side needs to be as smart as the market making community. If it were a poker game, you wouldn't show your hand, brokers shouldn't either if that is important to their clients orders.
This is correct.
ftobin
Posts: 1071
Joined: Fri Mar 20, 2009 3:28 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by ftobin »

gw wrote:Minor point: I think you're conflating RBC and IEX. Thor was developed by Katsuyama, the star of the Lewis book, and his crew while he was at RBC, and it was patented (somehow successfully) by RBC. Katsuyama later left RBC to create IEX. I don't know that IEX has rights to use Thor (maybe you do?), though it wouldn't be surprising.
I think you're correct about this. I don't see any reference to Thor on IEX's site. They do offer routing, but I don't see any commentary on it being a "better route". I know participants rest there, but not sure about the value-add in routing.
ftobin
Posts: 1071
Joined: Fri Mar 20, 2009 3:28 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by ftobin »

gw wrote:
IlliniSigEp wrote:I really am not trying to be Cpt. HFT, but I guess I will be for sake of discussion!

Jack's house analogy has some issues.

The situation would be more like this, there are 2 houses for sale, both for $100. You walk over to one house and buy it for $100. Your neighbor sees you do that, figures you might be interested in the other one, and runs out and buys it for $100. Now you have a choice, pay your neighbor more, or don't. Maybe you should look in another neighborhood for a $100 house! Or maybe, before you buy the first house, you get a buddy and walk to both house at the same time and buy both. No problem.
This is correct.
The analogy could be a little better in that you are actually approaching two houses simultaneously (you to one house and your spouse to another). One house is a little farther away. You purchase the first, closer. house, and someone else, sitting in front of the second house, sees the transaction, and then turns around and buys the second house.

Light is faster than your spouse's walking (or running!).
User avatar
black jack
Posts: 806
Joined: Fri Oct 10, 2008 10:13 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by black jack »

Ask the author: http://booktv.org/Program/15525/Author+ ... tquot.aspx
Michael Lewis, author of "Flash Boys," will appear LIVE on Book TV to answer questions about the book on Saturday, April 5th, at 11am ET. You can participate in the conversation by calling in during the program or by sending your questions/comments to Book TV via email (booktv@c-span.org), Facebook (facebook.com/BookTV), and Twitter (@BookTV). In "Flash Boys," Mr. Lewis looks at the hidden world of high-frequency stock trading and explains how high-frequency traders and big Wall Street banks have rigged the system to gain an advantage over regular investors. He also talks about a group of traders who, having discovered how the market now operates, created their own stock exchange to try to level the playing field.
We cannot absolutely prove [that they are wrong who say] that we have seen our best days. But so said all who came before us, and with just as much apparent reason. | -T. B. Macaulay (1800-1859)
manwithnoname
Posts: 1584
Joined: Mon Jul 22, 2013 7:52 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by manwithnoname »

ftobin wrote:
manwithnoname wrote:Jack Bogle has no problem with HFT.
People from the institutional world have only very, very slowly come to know what occurs in market microstructure.
So Michael Lewis knows more than Jack bogle about how stock markets work?

LOL
zeugmite
Posts: 1176
Joined: Tue Jul 22, 2008 11:48 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by zeugmite »

VictoriaF wrote:It's misleading to compare today's technological innovations to their yesterday's functional counterparts. Technology makes the difference. For example, if fMRI enabled us to read an investor's intention while it's still inside his brain, we would have to revise trading rules instead of insisting that we always had an inkling of what this investor is likely to do. Likewise, if modern computers and purchased proximity enable HFT traders to peek into Vanguard's "brain" while it's in the middle of a large transaction, new rules are required to reestablish the fairness of the market.
The core "issue" if it can be called that is access to information. A perfectly "fair" market tries to minimize information asymmetry between participants as much as possible, carrying out only reflections of individual wills over the same information rather than allowing additional gains based on better information. Any kind of asymmetry is inherently "insider trading," but practically speaking that can't be eliminated as markets are currently structured. For one, some people will be more proximate to the source of new information, and they are heavily incentivized to disseminate such information through market interactions rather than through other channels. Even if they were so kind to disseminate the information publicly, physics dictates that some people will receive the information sooner than others. Even if we allowed that all entities could receive new information simultaneously, different entities would still react with different speeds. None of this is "fair" as I've defined it above. HFT methods, due to newer technology, just exposed this truth that had always existed in markets formerly obscured by large latencies and inefficiencies, in an now entirely transparent and obvious way. The markets are and have been "rigged," structurally, not necessarily due to malfeasance.

How to get around it? Well, if the goal is to get everybody the same information, then a naive approach might be to try to get everybody the same access as HFT's, e.g. their low-delay analysis of data feeds and market probing results, or super fMRI readings of the future, should be published as the true market state, before they can trade on it. Of course then they'd have no incentives to figure all this out. A better approach is to do what computer systems have always done when faced with synchronization issues due to propagation delay -- latch and clock. Allow trades to cross only at quantized intervals of time across all exchanges simultaneously, and immediately publish all trades that happened. The interval should be no smaller than the current speed of non-market communication once around the world. This gives at least the theoretical possibility of information fairness.

tl;dr: Continuous trading is dumb and leads to fragmented unfair markets because physics.
manwithnoname
Posts: 1584
Joined: Mon Jul 22, 2013 7:52 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by manwithnoname »

zeugmite wrote:
VictoriaF wrote:It's misleading to compare today's technological innovations to their yesterday's functional counterparts. Technology makes the difference. For example, if fMRI enabled us to read an investor's intention while it's still inside his brain, we would have to revise trading rules instead of insisting that we always had an inkling of what this investor is likely to do. Likewise, if modern computers and purchased proximity enable HFT traders to peek into Vanguard's "brain" while it's in the middle of a large transaction, new rules are required to reestablish the fairness of the market.
The core "issue" if it can be called that is access to information. A perfectly "fair" market tries to minimize information asymmetry between participants as much as possible, carrying out only reflections of individual wills over the same information rather than allowing additional gains based on better information. Any kind of asymmetry is inherently "insider trading," but practically speaking that can't be eliminated as markets are currently structured. For one, some people will be more proximate to the source of new information, and they are heavily incentivized to disseminate such information through market interactions rather than through other channels. Even if they were so kind to disseminate the information publicly, physics dictates that some people will receive the information sooner than others. Even if we allowed that all entities could receive new information simultaneously, different entities would still react with different speeds. None of this is "fair" as I've defined it above. HFT methods, due to newer technology, just exposed this truth that had always existed in markets formerly obscured by large latencies and inefficiencies, in an now entirely transparent and obvious way. The markets are and have been "rigged," structurally, not necessarily due to malfeasance.

How to get around it? Well, if the goal is to get everybody the same information, then a naive approach might be to try to get everybody the same access as HFT's, e.g. their low-delay analysis of data feeds and market probing results, or super fMRI readings of the future, should be published as the true market state, before they can trade on it. Of course then they'd have no incentives to figure all this out. A better approach is to do what computer systems have always done when faced with synchronization issues due to propagation delay -- latch and clock. Allow trades to cross only at quantized intervals of time across all exchanges simultaneously, and immediately publish all trades that happened. The interval should be no smaller than the current speed of non-market communication once around the world. This gives at least the theoretical possibility of information fairness.

tl;dr: Continuous trading is dumb and leads to fragmented unfair markets because physics.
[OT comment removed by admin LadyGeek]
User avatar
Phineas J. Whoopee
Posts: 9675
Joined: Sun Dec 18, 2011 5:18 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by Phineas J. Whoopee »

Hi zeugmite,

I understand what you said, and follow your reasoning, but I'm having trouble relating it to the real world.

The proverbial Farmer in the Dell might occasionally be interested in the price of Greasy Wool futures, but usually is too busy tending sheep to watch for market moves and news. The futures market is interesting to her inasmuch as it enables said Dell Farmer to make profitable business decisions.

Greasy Wool futures Traders on the Wall, on the other hand, have nothing else to do, and watch like hawks.

The Dell and the Wall rely on each other for their mutual livelihood. The Wall offloads risk from the Dell, and the Dell offloads sheep from the Wall.

Even if everybody could access information symmetrically, they won't, because their needs are different.

And anyway, I stand by my earlier post that anyone can access and act on information quickly as long as they want to and are able to raise the, quite considerable, necessary investment capital.

PJW
Last edited by Phineas J. Whoopee on Fri Apr 04, 2014 7:02 pm, edited 1 time in total.
User avatar
VictoriaF
Posts: 20122
Joined: Tue Feb 27, 2007 6:27 am
Location: Black Swan Lake

Re: Help me understand Michael Lewis' "Flash Boys"

Post by VictoriaF »

manwithnoname wrote:
zeugmite wrote:How to get around it? Well, if the goal is to get everybody the same information, then a naive approach might be to try to get everybody the same access as HFT's, e.g. their low-delay analysis of data feeds and market probing results, or super fMRI readings of the future, should be published as the true market state, before they can trade on it. Of course then they'd have no incentives to figure all this out. A better approach is to do what computer systems have always done when faced with synchronization issues due to propagation delay -- latch and clock. Allow trades to cross only at quantized intervals of time across all exchanges simultaneously, and immediately publish all trades that happened. The interval should be no smaller than the current speed of non-market communication once around the world. This gives at least the theoretical possibility of information fairness.

tl;dr: Continuous trading is dumb and leads to fragmented unfair markets because physics.
Since it aint going to happen no need for further discussion..
If zeugmite's solution is technically feasible and would reduce the asymmetry of information, it's a useful discussion indeed.

Victoria
Inventor of the Bogleheads Secret Handshake | Winner of the 2015 Boglehead Contest. | Every joke has a bit of a joke. ... The rest is the truth. (Marat F)
User avatar
magellan
Posts: 3489
Joined: Fri Mar 09, 2007 3:12 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by magellan »

zeugmite wrote:A better approach is to do what computer systems have always done when faced with synchronization issues due to propagation delay -- latch and clock. Allow trades to cross only at quantized intervals of time across all exchanges simultaneously, and immediately publish all trades that happened. The interval should be no smaller than the current speed of non-market communication once around the world. This gives at least the theoretical possibility of information fairness.

tl;dr: Continuous trading is dumb and leads to fragmented unfair markets because physics.
I believe this solution is on the table currently and I'll go out on a limb and predict that it's a question of when, not if, it gets implemented.

In addition to solving the 'transparent same price everywhere' problem, a latch and clock design eliminates the incentive to cram the whole thing within a few miles of Wall Street. The result will be a more geographically dispersed system that's much more robust and resilient. IMO, it's almost reckless that such an important financial system is as geographically vulnerable as it currently is.

Jim
dl7848
Posts: 440
Joined: Tue Mar 25, 2014 12:46 am

Re: Help me understand Michael Lewis' "Flash Boys"

Post by dl7848 »

manwithnoname wrote:
ftobin wrote:
manwithnoname wrote:Jack Bogle has no problem with HFT.
People from the institutional world have only very, very slowly come to know what occurs in market microstructure.
So Michael Lewis knows more than Jack bogle about how stock markets work?

LOL
For Jack's sake, I hope he wasn't trying to conflate HFT with dark pools in the following quote. They are two separate things:
For instance, Bogle notes that high-frequency traders can benefit from getting an early peek at data on trading volume. So, he said, regulators need to push for more market transparency. Bogle added: "They're called dark pools for a reason."
User avatar
Topic Author
nisiprius
Advisory Board
Posts: 52216
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: Help me understand Michael Lewis' "Flash Boys"

Post by nisiprius »

magellan wrote:...cram the whole thing within a few miles of Wall Street...
One of the fascinating little things I hadn't quite realized was that glamorous Wall Street is really in unglamorous-sounding Mahwah, NJ, Weehauken, NJ, and Carteret, NJ.

Here's part of it, I think.

http://goo.gl/maps/eH59d

Image

No statues of charging bulls to be seen, but circled in red is the top-secret mammoth DB25 connector where the HFT traders plug in. joke...
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.
madbrain
Posts: 6809
Joined: Thu Jun 09, 2011 5:06 pm
Location: San Jose, California

Re: Help me understand Michael Lewis' "Flash Boys"

Post by madbrain »

nisiprius wrote:
magellan wrote:...cram the whole thing within a few miles of Wall Street...
One of the fascinating little things I hadn't quite realized was that glamorous Wall Street is really in unglamorous-sounding Mahwah, NJ, Weehauken, NJ, and Carteret, NJ.

Here's part of it, I think.

http://goo.gl/maps/eH59d

Image

No statues of charging bulls to be seen, but circled in red is the top-secret mammoth DB25 connector where the HFT traders plug in. joke...
LOL . I counted 38 trees, and they are not really in the proper DB connector shape .
madbrain
Posts: 6809
Joined: Thu Jun 09, 2011 5:06 pm
Location: San Jose, California

Re: Help me understand Michael Lewis' "Flash Boys"

Post by madbrain »

manwithnoname wrote:
ftobin wrote:
manwithnoname wrote:Jack Bogle has no problem with HFT.
People from the institutional world have only very, very slowly come to know what occurs in market microstructure.
So Michael Lewis knows more than Jack bogle about how stock markets work?

LOL
Jack Bogle hasn't worked in the industry for quite a while, has he ?
Was HFT even around when he left Vanguard ?

He is a great man, but I don't think he is necessarily the final authority on HFT.
I haven't read the book yet either. I probably will. It seems that HFT is awfully complicated.

The issue about how much HFT skims, and whether there are really benefits for these costs, is an interesting thing.
But I what I really want to know about is if we understand how the past flash crashes happened, whether HFT caused them, and what can be done about them.
Iorek
Posts: 1569
Joined: Fri Mar 08, 2013 8:38 am

Re: Help me understand Michael Lewis' "Flash Boys"

Post by Iorek »

manwithnoname wrote:
So? There is nothing of substance to discuss. SEC and FINRA are aware of what is going on. Until they take action their inactivity indicates that they have no objections to current trading practice under Reg NMS. No brainer.
Is your position really that if regulatory agencies have not taken action then there is no possibility-- and no point in discussing the possibility-- that certain market practices may be adverse to individual investors or the market generally?

No brainer indeed.
User avatar
in_reality
Posts: 4529
Joined: Fri Jul 12, 2013 6:13 am

Re: Help me understand Michael Lewis' "Flash Boys"

Post by in_reality »

madbrain wrote:
nisiprius wrote:
magellan wrote:...cram the whole thing within a few miles of Wall Street...
One of the fascinating little things I hadn't quite realized was that glamorous Wall Street is really in unglamorous-sounding Mahwah, NJ, Weehauken, NJ, and Carteret, NJ.

Here's part of it, I think.

http://goo.gl/maps/eH59d

Image

No statues of charging bulls to be seen, but circled in red is the top-secret mammoth DB25 connector where the HFT traders plug in. joke...
LOL . I counted 38 trees, and they are not really in the proper DB connector shape .
Obviously it's a non-standard variant of a D-SUB 37 connectors commonly used in Hospital facilities as an interface between hospital beds & nurse call systems, allowing for the connection & signaling of Nurse Call, Bed Exit, and Cord out including TV entertainment & lighting controls.

Think about it, they can get kickbacks from the manufacturer who obviously can overcharge for it.
ftobin
Posts: 1071
Joined: Fri Mar 20, 2009 3:28 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by ftobin »

zeugmite wrote:Allow trades to cross only at quantized intervals of time across all exchanges simultaneously, and immediately publish all trades that happened.
I like your term "quantized intervals" :)

For everyone's knowledge, the usage of a "cross" (in contrast to continuous trading) is something that has increased over time.

In the beginning (I'll use NASDAQ), there were no aggregate crosses. Everything was continuous. Prices for the open and close were widly manipulated. Oh, the games that were played before opening cross. Eventually they came up with the NASDAQ Opening Cross (2004 or 2005), which allowed the opening print to become less volatile. It aggregated all premarket interest and crossed it at the price that transacted the most trades.

After NASDAQ Opening Cross, they applied to to the close, which became the NASDAQ Closing Cross. Things stayed like this for a few years until ARCA began listing stocks, and developed its own crossing mechanism. The NYSE, all this time, had it's specialists handling the crosses. NYSE's crosses are fairly opaque. NASDAQ's and ARCA's are transparent and algorithmically determined.

For a couple years, we just had opening and closing crosses on the major exchanges. Then, intraday single-stock circuit breakers allowed stocks to temporarily halt during the day. When they unhalted, the exchange would use a crossing mechanism similar to the open/close. This happened fairly infrequently.

Just last year, Limit Up/Limit Down went into effect, further creating better safeguards to protect against major unchecked movements in a stock (can't move more than a certain percentage in a rolling 15-minute window). Today, stocks temporarily halt all the time from LULD for 5 minutes, and the exchanges, again, use a cross just like during the open and close. It helps reduce volatility and fairly aggregates interest during the holding period.

The solution zeugmite is describing is something further extends usage of crosses to all regular trading. It works easily with one exchange, but it would be challenging to implement in the US, since we currently have several exchanges and ECNS in the US that all trade independently (NYSE, NASDAQ, ARCA, BATS, EDGX, EDGA, BYX, BX, FLOW, and more). Questions abound. How is interest aggregated? Does each exchange have its own cross? If so, that's going to cause problems, since you no longer have an NBBO (national best bid & offer) protecting you from being executed outside of it. Cross execution prices would jump around from exchange to exchange in the same interval. The entire rebate/charge pricing model which drives the exchanges would change. Do darkpools interact? Etc., etc.

If someone can devise how to handle the fragmentation, I'd truly like to hear it.
adios_logic
Posts: 22
Joined: Mon Aug 05, 2013 8:54 am

Re: Help me understand Michael Lewis' "Flash Boys"

Post by adios_logic »

I don't really know where to start with how off target this thread is. I guess it starts with a question that assumes that Michael Lewis "is telling the truth and is accurate" which is a bad place to start. In any event, I will disclose that I have a dog in this fight. But that's not a reason for you not to think about the arguments (Michael Lewis and Brad from his book also have dogs in the fight).

High frequency traders, market makers, liquidity providers, whatever name you want to call them enter bids and offers by using publicly available market data that likely includes information about supply and demand in that stock, other related stocks, relevant futures, foreign stocks, etc. They process the information and use various forms of statistical analysis to determine what they think is a fair price. They adjust that price for the risk they are taking. HFT firms have automated tools that allow them to manage risk efficiently compared to the liquidity providers from 10-20 years ago, so it lets them adjust less for the risk they are taking. So if you determined that MSFT is worth 39.87449, you might be willing to buy at 39.87 and sell at 39.88. You would probably also have bids outstanding at lower prices and offers at higher prices. There are several exchanges in the US, so you might do this across some (but probably not all of them). Just noting that ETFs are different in that you can calculate their price based on the underlying securities. Who do you think does that? High frequency traders. The tools they use to trade make it natural that they would be good at this. This allows them to make sure they keep ETFs really close to fair value even if they aren't liquid.

Now remember, HFT/market makers/liquidity providers role in the market is to stand willing to buy and sell and to help balance supply and demand to find the market clearing price. So now we are left with two explanations of why the price would move when someone starts buying a lot of stock.

1. Their buying changes the supply and demand dynamics and the HFT/MMer/liquidity providers respond and change the market clearing price -- meaning they play their role in the market, or
2. They magically divine that 100 share trade is actually the first 100 shares of a 50,000 share order and adjust all their prices and maybe even race to the other markets to lift all the offers to sell the stock back to that unwitting buyer.

With respect to 1, which is the obvious choice. 2 is an unrealistic, overly simplistic and naive strategy and basically what they claim in the book. If the buyer does a bad job, they can leak information and tip their hand more than necessary (note that RBC appeared to be doing an unbelievable bad job for a professional trading firm). The idea of "Thor" was to do a good job and not leak any unnecessary information. Note that I say "unnecessary" because trades are public, so trading releases information that everyone else responds to. This is a good thing -- it allows the market to find the new, right price.

So pretend you were a liquidity provider and that all the scary speed is slowed down to human speed. Let's say there are 8 exchanges all with 5,000 shares to sell at 39.88 and your order is one of them - so 40,000 shares total. Then you notice that one exchange's quote is executed, then the 2nd, then the 3rd (remember you are trying to make money doing this). You might start thinking that the price is likely to go up and if you sell at 39.88, you are going to lose money. Then you notice the 4th exchange's 5,000 shares get executed. Then you see the 6th exchange update their price to 39.89. If I were you, I would update my price, the supply and demand dynamics have changed. You would be crazy not to update your price. I don't think you would be doing something wrong by updating your price... your job is to help find the right price, not lose money.

The idea of Thor was to try to trade on all 8 exchanges at about the same time. This allowed them to avoid leaking the trade information of each exchange that allowed you and others to update your prices. The idea behind Thor is obvious and other people were already doing it... because it was obvious. There was even a name for it -- spraying.

So speed that up to computer time and that's what happens. But remember, the HFT/MM/liquidity provider isn't just pricing based on that stock, so they might have noticed that S&P Futures ticked down or a related stock ticked down and they think that 39.88 is a great price. Then they will trade at 39.88 and everyone is happy. If you want to watch this super long, but actually reasonable debate/conversation about HFT (http://insider.thomsonreuters.com/link. ... MDJhN2I%3D) a high frequency trader says he makes money on 53% of his trades. That means he is wrong a lot of the time. But he trades a lot, so over the course of hundreds of thousands of trades, he is profitable.

I know this was super long, but it's painful for me to read conspiracy theories when there are logical, obvious answers. It sucks that Michael Lewis took this guy Brad at his word instead of thinking about it. But it's strange to me that bogleheads would take a writer's word over Vanguard's (http://www.bloomberg.com/news/2014-04-0 ... abuse.html) and Jack Bogle (http://www.cbsnews.com/news/jack-bogle- ... d-markets/). Clifford Asness from AQR also wrote a good piece: http://online.wsj.com/news/articles/SB1 ... 2237652362.

Finally this piece clears up some easy stuff that was obviously wrong in the book: http://www.businessweek.com/articles/20 ... trading#p1
ftobin
Posts: 1071
Joined: Fri Mar 20, 2009 3:28 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by ftobin »

adios_logic wrote:The idea of Thor was to try to trade on all 8 exchanges at about the same time. This allowed them to avoid leaking the trade information of each exchange that allowed you and others to update your prices. The idea behind Thor is obvious and other people were already doing it... because it was obvious. There was even a name for it -- spraying.
First, I'll state I agree with your examples (I don't see the calculated precision of MSFT being out to 4 places, but whatever :-) )

The key with Thor is that it's a value-added spray. They have to time each outgoing order so that it arrives at each exchange simultaneously. I've got no problem any with this.

The only issues I like to highlight is that (1) what Thor is doing is patented, so it's technically illegal to do yourself and (2) the huge capital investment takers have to go to do something as simple as execute against all the posted bright markets successfully.

It's arguable whether or not what is happening is against regulation (SEC or another SRO could re-interpret its rules and apply them against certain behaviors). The question is more whether or not we want to have this behavior. It's kind similar to before insider trading rules were established. It would have been fully legal then, but we decided we didn't like the incentives/outcomes and changed things.

Surprisingly, the Pinksheet market operates more fairly in this regard. Pinksheets locks in the target MM's obligation when your order is placed.
SBritt
Posts: 38
Joined: Sun Aug 25, 2013 11:36 am

Re: Help me understand Michael Lewis' "Flash Boys"

Post by SBritt »

Here's a good YouTube video about HFT - It may not explain it as detailed as you like, but it does help you understand it at a broad level. These guys that design the systems are truly some brilliant minds.

http://youtu.be/GEAGdwHXfLQ
adios_logic
Posts: 22
Joined: Mon Aug 05, 2013 8:54 am

Re: Help me understand Michael Lewis' "Flash Boys"

Post by adios_logic »

The key with Thor is that it's a value-added spray. They have to time each outgoing order so that it arrives at each exchange simultaneously. I've got no problem any with this.
I have no problem with Thor taking everything out at once... they should do that and the firms posting should always be willing to trade if they posted quotes. I think it's a bit over-engineered, but if it works, that's great.
The only issues I like to highlight is that (1) what Thor is doing is patented, so it's technically illegal to do yourself and (2) the huge capital investment takers have to go to do something as simple as execute against all the posted bright markets successfully.
On 2, I agree, but this is an issue of market design... intermediaries like HFT/MMers/Liquidity providers help to keep our super complex market functioning. Would we be better off if the market wasn't so complex? Yes. But the notion that HFT is endemic to US Equities is silly. In fact, there are a lot of markets that are monopolies and their products only trade on one market, so none of the craziness in the book could possibly happen, and guess what? The same firms trade them just as effectively and it is nowhere near as expensive for anyone. Kind of sad that our equities market has such a high barrier to entry. Lots of firms commented against Reg NMS, it had noble intentions, but our markets could be a lot better. Wouldn't it be nice if you could explain how the market works to your kids?
It's arguable whether or not what is happening is against regulation (SEC or another SRO could re-interpret its rules and apply them against certain behaviors).
I mostly disagree with this, but it wouldn't be the first time they did that. Acting on public information, even if you process it quickly is legal (there's a difference between seeing the future and seeing the past quickly). That's not to say that no one uses a computer to do something illegal in the market. Obviously regulators should find them, stop the behavior, fine them, and so on.
The question is more whether or not we want to have this behavior.
Exactly. Some of the behavior we effectively encourage doesn't add any value. When your markets are overly complex and not transparent enough, people find a way to exploit inefficiencies and that's bad. The best way to solve that is to make them more straightforward and more transparent. But regulators always add new rules making it more complex, making matters worse. For example, if we didn't have rules that subsidize exchanges, perhaps we wouldn't have so many and we would see less of this behavior. Do you think that the smallest several exchanges on this list are adding or extracting value? Brokers basically have to connect and pay their membership and connection fees under Reg NMS. http://batstrading.com/market_summary/
kayo
Posts: 60
Joined: Sat Aug 18, 2012 9:54 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by kayo »

An editorial by Joe Nocera about the Michael Lewis book is now on nytimes.com:

http://www.nytimes.com/2014/04/05/opini ... rusade.htm
ftobin
Posts: 1071
Joined: Fri Mar 20, 2009 3:28 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by ftobin »

adios_logic wrote:Do you think that the smallest several exchanges on this list are adding or extracting value? Brokers basically have to connect and pay their membership and connection fees under Reg NMS. http://batstrading.com/market_summary/
To be fair, with the smaller exchanges, they normally don't have liquidity, and you'll just use annother exchange's router to access the oddballs. It's sillly for everyone to connect to the Chicago SE and such :-)
gw
Posts: 776
Joined: Thu May 28, 2009 10:12 am

Re: Help me understand Michael Lewis' "Flash Boys"

Post by gw »

ftobin wrote:
adios_logic wrote:Do you think that the smallest several exchanges on this list are adding or extracting value? Brokers basically have to connect and pay their membership and connection fees under Reg NMS. http://batstrading.com/market_summary/
To be fair, with the smaller exchanges, they normally don't have liquidity, and you'll just use annother exchange's router to access the oddballs. It's sillly for everyone to connect to the Chicago SE and such :-)
I disagree. In practice, I'm handicapped unless I have access to all RegNMS-protected venues. For example, if I want to trade immediately against both the best and 2nd-best offer in the market, then before I can trade against the 2nd-best offer at Nasdaq, RegNMS requires me to assure Nasdaq that I've already routed to any offer Chicago has at the NBBO.
User avatar
magellan
Posts: 3489
Joined: Fri Mar 09, 2007 3:12 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by magellan »

ftobin wrote:The solution zeugmite is describing is something further extends usage of crosses to all regular trading. It works easily with one exchange, but it would be challenging to implement in the US, since we currently have several exchanges and ECNS in the US that all trade independently (NYSE, NASDAQ, ARCA, BATS, EDGX, EDGA, BYX, BX, FLOW, and more). Questions abound. How is interest aggregated? Does each exchange have its own cross? If so, that's going to cause problems, since you no longer have an NBBO (national best bid & offer) protecting you from being executed outside of it. Cross execution prices would jump around from exchange to exchange in the same interval. The entire rebate/charge pricing model which drives the exchanges would change. Do darkpools interact? Etc., etc.

If someone can devise how to handle the fragmentation, I'd truly like to hear it.
As always, the devil is in the details, but I don't think the synchronization needs to be as complicated as you describe. The central idea is that every exchange is synchronized to the same highly accurate clock signal. The clock signal is a square wave with equally spaced rising and falling edges. Trades can only happen on a rising or falling edge of the signal. The clock edges are far enough apart that there's enough time for all of the exchanges to publish everything that happened at the last clock edge and for traders to decipher the info and set up the next round of trades. There would of course be other details, but that's the crux of it.

There are other examples of globally synchronized geographically dispersed systems that work this way. GPS is a great example. In order for your GPS to be able to tell where you are, it needs to know exactly what time it is, down to I think around 40ns. A GPS receiver uses the time it took to receive a satellite signal, sort of a propagation delay, to determine exactly how far away the satellite is. The receiver repeats this timing operation against three or more satellites and combines this segment length information with known coordinates for each satellite to triangulate your position. The accuracy of the position is largely a function of the accuracy of the gps receiver's synchronized clock.

Jim
Last edited by magellan on Sat Apr 05, 2014 8:22 am, edited 3 times in total.
User avatar
alec
Posts: 3181
Joined: Fri Mar 02, 2007 1:15 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by alec »

gw wrote:
ftobin wrote:
adios_logic wrote:Do you think that the smallest several exchanges on this list are adding or extracting value? Brokers basically have to connect and pay their membership and connection fees under Reg NMS. http://batstrading.com/market_summary/
To be fair, with the smaller exchanges, they normally don't have liquidity, and you'll just use annother exchange's router to access the oddballs. It's sillly for everyone to connect to the Chicago SE and such :-)
I disagree. In practice, I'm handicapped unless I have access to all RegNMS-protected venues. For example, if I want to trade immediately against both the best and 2nd-best offer in the market, then before I can trade against the 2nd-best offer at Nasdaq, RegNMS requires me to assure Nasdaq that I've already routed to any offer Chicago has at the NBBO.
I think that ftobin is saying that you don't have to be a member of all the exchanges to route orders to those exchanges. A broker can send an order to the exchange at which it's a member, directing it's exchange to route the order (through that exchange's outbound router) to an exchange at which it's not a member. This is just using a member of another exchange to access that exchange.

Aha, found NASDAQ's NMS order types.
Last edited by alec on Sat Apr 05, 2014 7:39 am, edited 1 time in total.
"It is difficult to get a man to understand something, when his salary depends upon his not understanding it!" - Upton Sinclair
User avatar
alec
Posts: 3181
Joined: Fri Mar 02, 2007 1:15 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by alec »

ftobin wrote:
Surprisingly, the Pinksheet market operates more fairly in this regard. Pinksheets locks in the target MM's obligation when your order is placed.
ftobin,

Can you explain a little more?

thanks
"It is difficult to get a man to understand something, when his salary depends upon his not understanding it!" - Upton Sinclair
adios_logic
Posts: 22
Joined: Mon Aug 05, 2013 8:54 am

Re: Help me understand Michael Lewis' "Flash Boys"

Post by adios_logic »

gw wrote:
ftobin wrote:
adios_logic wrote:
Do you think that the smallest several exchanges on this list are adding or extracting value? Brokers basically have to connect and pay their membership and connection fees under Reg NMS. http://batstrading.com/market_summary/

To be fair, with the smaller exchanges, they normally don't have liquidity, and you'll just use annother exchange's router to access the oddballs. It's sillly for everyone to connect to the Chicago SE and such :-)


I disagree. In practice, I'm handicapped unless I have access to all RegNMS-protected venues. For example, if I want to trade immediately against both the best and 2nd-best offer in the market, then before I can trade against the 2nd-best offer at Nasdaq, RegNMS requires me to assure Nasdaq that I've already routed to any offer Chicago has at the NBBO.


I think that ftobin is saying that you don't have to be a member of all the exchanges to route orders to those exchanges. A broker can send an order to the exchange at which it's a member, directing it's exchange to route the order (through that exchange's outbound router) to an exchange at which it's not a member. This is just using a member of another exchange to access that exchange.
First, I apologize, but you are far beyond me with your quote embedding skills. My point was that your broker has to connect (or more likely the wholesaler your broker sells your orders to -- sells not sends, well both sells and send, I guess). But that isn't free to you. That's part of the cost. It's really small, but I'm cheap and I don't want to pay an extra cent if I don't have to. In any event, If we let market forces determine how many exchanges there were, the exchanges that extract value would go away. Reg NMS forces those brokers to connect giving them one revenue stream and gives them a big revenue stream from the $300 million or more of market data revenue from the SIP that probably costs around $10m to operate.

That is not a good link for NMS order types: here is nasdaq's presentation: http://www.brainshark.com/nasdaqomx/vu? ... 6aUZzoG0z0
If you don't want to click on that seemingly sketchy link go here: http://nasdaqtrader.com/Trader.aspx?id= ... USEquities and click to view the presentation.

Here is BATS's: http://batstrading.com/features/ with links to their order types and DirectEdge's (which they own)
ARCA's: http://usequities.nyx.com/markets/nyse- ... rder-types

If you actually looked at all those, your reaction is probably, "What the?" Yes, it's super complex. And those are summaries, look at the technical specs that programmers use to code to the exchanges -- you end up with interactions where you can use order types in combination. You can see why Brad had no idea what was going on. I put order types into 3 categories: 1) stuff exchanges do to comply with NMS, 2) automating simple things traders used to do (pegging and discretion, etc, and 3) other stuff. I don't really count routing as order types, but find, 4) routing strategies.

1 would mostly go away if we reformed NMS -- these are the most complicated. 2 and 4 are useful. 3 is a mixed bag. Some are debatable that they add value and others are probably great.
adios_logic
Posts: 22
Joined: Mon Aug 05, 2013 8:54 am

Re: Help me understand Michael Lewis' "Flash Boys"

Post by adios_logic »

Here's a question that I don't think anyone has addressed, why does a 350 microsecond delay for all users protect people for HFT tactics? If you slow everyone down by the same amount, can't HFTs still beat you to another market? If the guy shooting the gun to start a race waits an extra millisecond, it doesn't matter, everyone still starts at the same time so the fastest racer wins. Is everyone really taking this on faith?
User avatar
magellan
Posts: 3489
Joined: Fri Mar 09, 2007 3:12 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by magellan »

adios_logic wrote:Here's a question that I don't think anyone has addressed, why does a 350 microsecond delay for all users protect people for HFT tactics? If you slow everyone down by the same amount, can't HFTs still beat you to another market? If the guy shooting the gun to start a race waits an extra millisecond, it doesn't matter, everyone still starts at the same time so the fastest racer wins. Is everyone really taking this on faith?
The idea is not to delay trading, but to synchronize it. Did you see my post above?

Jim
HoosierJim
Posts: 987
Joined: Wed Mar 24, 2010 7:11 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by HoosierJim »

ftobin wrote: They don't have to actually place the 100-share offers, just listen to exchange A's data feed to pull the quotes on other exchanges.
There is actually an advantage of chumming with the small quantity share offer and letting it complete. The order execution message occurs BEFORE the update feed (but only to trader X). So if trader X's order gets executed at 100.00, trader X knows the price earlier than others that are just waiting for the feed update. This gives trader X a slight information price & time advantage which is X's reward for liquidity. X now has an indication of market movement vs the "lurkers". If trader X has faster links to other exchanges (i.e. microwave vs fiber) - trader X can develop a strategy and beat others to the other exchanges.

Here's an idea http://faculty.chicagobooth.edu/eric.bu ... ctions.pdf
adios_logic
Posts: 22
Joined: Mon Aug 05, 2013 8:54 am

Re: Help me understand Michael Lewis' "Flash Boys"

Post by adios_logic »

adios_logic wrote:
Here's a question that I don't think anyone has addressed, why does a 350 microsecond delay for all users protect people for HFT tactics? If you slow everyone down by the same amount, can't HFTs still beat you to another market? If the guy shooting the gun to start a race waits an extra millisecond, it doesn't matter, everyone still starts at the same time so the fastest racer wins. Is everyone really taking this on faith?
The idea is not to delay trading, but to synchronize it. Did you see my post above?
I was referring to IEX's 350 microsecond delay. The batch auction stuff seems like a mess. You aren't going to synchronize exchanges operating in different places. So you would need a monopoly exchange. Plus the information that affects prices don't happen in discrete intervals, so it simply doesn't solve for information advantage. If I have information from a continuous source and you don't when an auction starts, I will do the right thing and you may not. I believe that the more artificial delays and barriers you put in place and the less transparent a market is, the more prone to abuse it is. Not the opposite. But this is America... someone should start the batch auction exchange and we'll see what happens. I'm skeptical of things the government would have to mandate to work.
Jeff Albertson
Posts: 904
Joined: Sat Apr 06, 2013 7:11 pm
Location: Springfield

Re: Help me understand Michael Lewis' "Flash Boys"

Post by Jeff Albertson »

Michael Lewis discussed the book on KQED radio yesterday. The 52 minute interview is available for download from the KQED web site:

http://www.kqed.org/a/forum/R201404041000
zeugmite
Posts: 1176
Joined: Tue Jul 22, 2008 11:48 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by zeugmite »

adios_logic wrote:I was referring to IEX's 350 microsecond delay. The batch auction stuff seems like a mess. You aren't going to synchronize exchanges operating in different places. So you would need a monopoly exchange. Plus the information that affects prices don't happen in discrete intervals, so it simply doesn't solve for information advantage. If I have information from a continuous source and you don't when an auction starts, I will do the right thing and you may not. I believe that the more artificial delays and barriers you put in place and the less transparent a market is, the more prone to abuse it is. Not the opposite. But this is America... someone should start the batch auction exchange and we'll see what happens. I'm skeptical of things the government would have to mandate to work.
Well, everybody has information from a continuous source, i.e. the world, and you are encouraged to do the right thing and bring that information into the market. Because everybody can observe the world, sometimes you win sometimes you lose depending on where things happen. But I'm not talking about that. I'm talking about order flow information already present on the exchanges in what people consider as a "single market." If it is really a single market, then the market state should be synchronized. There should be no latency arbitrage or other continuous-trading timing inefficiencies at all. So the real question to ask yourself is, why wouldn't you want a single market? I can think of some reasons but they are rather esoteric; I am also pretty sure they are not the ones that entities currently reaping the benefit of market fragmentation are thinking. The rest are merely technical problems and technical problems have technical solutions if you want to find them.

It's not clear to me what IEX is doing. It's one design that may solve some problems, or maybe not, who knows.
User avatar
magellan
Posts: 3489
Joined: Fri Mar 09, 2007 3:12 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by magellan »

adios_logic wrote:The batch auction stuff seems like a mess. You aren't going to synchronize exchanges operating in different places. So you would need a monopoly exchange.
Yes, you can synchronize exchanges operating in different places, and no, you don't need a monopoly exchange. Exchanges already operate under a very complex regulatory framework. Introducing synchronization would probably reduce the complexity of the regulatory framework in some areas.

There's nothing technologically challenging or ground-breaking here. Well established synchronization approaches are taught to engineers in their very first digital logic design course. Potential problem areas have been identified and studied to death and there's a rich vocabulary of terms including things like propagation delay, setup time, hold time, clock skew and jitter. These all transfer in a pretty straightforward way to the trading problem. Global time synchronization adds a touch of extra complexity, but again, designs that do this exact thing have been widely deployed since at least the 1980s.

Jim
adios_logic
Posts: 22
Joined: Mon Aug 05, 2013 8:54 am

Re: Help me understand Michael Lewis' "Flash Boys"

Post by adios_logic »

Well, I can think of one big one, not much else. It would be a monopoly and that would be very likely to lead it to be not very good and very expensive. Generally consumers prefer competition. I agree there are pluses and minuses to it. But if you had one market, why not make it continuous? The latency stuff could be solved by putting the markets all in one data center. That would save everyone tons of money because they wouldn't pay for all the lines to connect all the data centers, but again has its own likely issues.
ftobin
Posts: 1071
Joined: Fri Mar 20, 2009 3:28 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by ftobin »

magellan wrote:There's nothing technologically challenging or ground-breaking here. Well established synchronization approaches are taught to engineers in their very first digital logic design course. Potential problem areas have been identified and studied to death and there's a rich vocabulary of terms including things like propagation delay, setup time, hold time, clock skew and jitter. These all transfer in a pretty straightforward way to the trading problem. Global time synchronization adds a touch of extra complexity, but again, designs that do this exact thing have been widely deployed since at least the 1980s.
I don't see synchornization as the problem. It's trival to get each exchange to cross trades at the same moment. The problem is how to handle disparities in imbalances between exchanges on each time slice.

Example: For one time interval, NASDAQ's imbalance is going to cross at 10.05. BATS, having fewer sellers at the same moment, is cross at 10.30. EDGX's book, having few buyers, is going to cross at 9.50. Basically, we're re-introduced a problem RegNMS is was meant to solve: banding where trades can execute (in particular, by the national best bid/offer -- NBBO). Each exchange underneath our current continuous trading paradigm bands where trades can execute -- your resting limit order can't be traded through. If we're using periodic, synchroniized crosses, we don't have this guarantee that your limit order will execute -- trades can happen through it.
User avatar
ogd
Posts: 4876
Joined: Thu Jun 14, 2012 11:43 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by ogd »

magellan wrote:
zeugmite wrote:A better approach is to do what computer systems have always done when faced with synchronization issues due to propagation delay -- latch and clock. Allow trades to cross only at quantized intervals of time across all exchanges simultaneously, and immediately publish all trades that happened. The interval should be no smaller than the current speed of non-market communication once around the world. This gives at least the theoretical possibility of information fairness.

tl;dr: Continuous trading is dumb and leads to fragmented unfair markets because physics.
I believe this solution is on the table currently and I'll go out on a limb and predict that it's a question of when, not if, it gets implemented.

In addition to solving the 'transparent same price everywhere' problem, a latch and clock design eliminates the incentive to cram the whole thing within a few miles of Wall Street. The result will be a more geographically dispersed system that's much more robust and resilient. IMO, it's almost reckless that such an important financial system is as geographically vulnerable as it currently is.

Jim
I love the computer clock analogy. Indeed, what I remember from digital circuit design classes is how quickly the simplest clock-less design becomes a hopeless mess, and how many advantages you get from having a stable state -- the ability to capture and reproduce problems, among other things. Doesn't it sound like a useful thing for markets?

For some of the challenges involved, see this old article: http://www.technologyreview.com/feature ... ess-chips/ . Choice quote:
And we all benefit: the discipline of clock-based design has enabled the magic of exponential growth in chip performance to endure for more than 30 years. “The clock has to go down as one of the most brilliant ideas in design,” says Kevin Normoyle, a Distinguished Engineer at Sun who works on the design of Sun’s Sparc microprocessors.
Spoiler alert: clock-less chips still haven't happened 13 years later, despite the relentless drive for speed. Which is, btw, something we don't really need in the markets above current levels, but we sure could use the discipline and ability to understand what's happening.
ftobin
Posts: 1071
Joined: Fri Mar 20, 2009 3:28 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by ftobin »

alec wrote:
ftobin wrote:Surprisingly, the Pinksheet market operates more fairly in this regard. Pinksheets locks in the target MM's obligation when your order is placed.
Can you explain a little more?
In Pinksheets, for the most part, exchanges don't play a part. Each market maker routes orders to other makers through a central system, directing system. Quoting, also happens a central system, which is tied to the order placement system. This central system is run by OTC Markets. If you see GSCO and UBSS quoting on the offer, and send an order to either/both of them, OTC Markets knows of both the order and their current quote, and can tie this information together.

Let's say you place an order to both GSCO and UBSS to buy from their quoted offer. When OTC Markets receives the order, they note if you're the first participant in to the target maker maker, and give you that back that information on each order acknowledgement. Now, note that the rules state that if you're the first person accessing a quote, the market maker must execute you. Not doing so is called backing away from a quote, and is a firm no-no.

So, if you send an order to both GSCO and UBSS simultaneously and you're the first one to do so, OTC Markets will tell you so. If one of them changes their quote in response to seeing the other market maker's execution, and changes/cancels their quote without giving you an execution, you have the right to follow-up and demand an execution for the quote size.

Now, I'll qualify this as all this is how it's supposed to work, if market makers were called out whenever they didn't follow the rules. However, they back away from quotes all the time. If they did follow the rules, it's technically a fairer system -- you're guaranteed to access the bright quotes you saw.

Note that the Pinksheets world can do this because it has a centralized player (for the most part). Our NMS stocks (NASDAQ/NYSE/ARCA-listed) have many more "centralized" systems (exchanges).
ftobin
Posts: 1071
Joined: Fri Mar 20, 2009 3:28 pm

Re: Help me understand Michael Lewis' "Flash Boys"

Post by ftobin »

adios_logic wrote:I was referring to IEX's 350 microsecond delay.
Jargon-filled response: The delay is to allow for IEX's NBBO to update so that resting passive peglims don't end up with effectively midspread/aggressive executions from incoming snipes on an NBBO move.
Post Reply