Friday, 22 February 2019

IEX - the cable guy

IEX has a pretty, petty, and embarrassing youtube advert that is kind of funny in a cringeworthy way:

The basis for the ad is to build on their junk whitepaper.  IEX improperly represents technology costs at other exchanges.

Far from being the fairest girl in Mordor, IEX is the evil one plumbing the darkest depths looking to give you an unpleasant burn around an unmentionable ring. Ramble on.

IEX is complaining about business models here.

For example, Aquis, the exchange with a dodgy and inefficient execution model, only charges for the cable. Aquis would argue that is fairer than IEX not charging for a cable. While IEX doesn't charge for their market data, IEX has very expensive trading fees, baked in unfairness, and a poor execution model. At IEX when you pay nothing for your cable, perhaps you're getting what you pay for. At Aquis, you can trade all you like for free if their dubious model suits your poor taste. Who is better there? Aquis is probably a better place to trade than IEX just on the basis that IEX makes lots of money from very little trading with their high fees and the Aquis income line is fairly lightweight. At the end of the day, both exchanges' weak models impair market structure. The cost of impairing economic efficiency should be better understood by regulators.

BATS also used to have a free market data model, but they succumbed to the pressure of profitability by eventually adding fees for that too. ICE/NYSE, CBOE, and Nasdaq certainly charge some big fees for their fast market data. When you spend a few hundred million dollars on a co-location facility in Mahwah, plus a suped-up Ferrari to run your low-latency high-bandwidth network, you end up with slightly more costs to reclaim than IEX crudely and rudely ruminates on.

Most exchanges try to balance the fees between trading and market data. Market data, like petroleum for your car, is pretty inelastic in its demand curve. Price whatever: you have to have it, or you can't get your work done. Reg NMS, as the highway system for trades, is powered by market data. High market data fees act as a tax on efficiency.

Trading fees are more elastic in demand than market data. As you seek out best execution, you don't have to trade everywhere. You just have to avoid trading poorly. IEX's high trading fees represent an unfortunate tax on efficiency. At least, you're not forced to trade on IEX's dodgy two-bit exchange even if you need to consume their dodgy data feed directly or indirectly.

NYSE and Nasdaq have had high rates of growth in market data fees. Many think this is unfair, including SIFMA,
"This pragmatic ruling by the SEC indicates increasing recognition by policymakers that the fee structure for proprietary market data products is broken. As noted in the unanimous decision, ‘the exchanges fail to meet their burden to demonstrate that the fees are fair and reasonable and not unreasonably discriminatory’ as required under current law. Today’s decision should prompt further examination of policy reforms to ensure the efficiency of public market data feeds and fairness of fees"
The SEC recognised this,
"Today, the Commission held that neither NYSE Arca, Inc. (“Arca”) nor Nasdaq Stock Market LLC (“Nasdaq”) had met its burden to show that the fees, filed in 2010 for ArcaBook and Level 2 (together, the “products”) and subsequently challenged by SIFMA, were fair and reasonable under the Securities Exchange Act of 1934."
The SEC paused this silliness. This marketing BS from IEX has at least now risen from toxic misrepresentation by whitepaper, to being kind of funny in a cringeworthy way. Perhaps by lowering the debate, IEX's schmarketing is at least more palatable. Hence the greater danger in misrepresentation here. More people may digest this without the necessary degree of cynicism. This diarrhoea for your grey matter is undeserving of grey water.

I don't think this kind of advertising could stand in Australia. We have pretty tight laws about false and misleading advertising. Many US adverts I see on US television wouldn't pass muster here. I guess IEX will do anything to try and get to a monthly 3%. Ethics begone.

This is not so dissimilar to their false and misleading claims regarding long and short queues at exchanges where IEX improperly dropped CHX's short queues from their charts. It's also a bit similar to how they project mid-point trading to being good when it misses the point that trading at best gains you half a spread over mid-point. Their rebates are called discounts to hide their hypocrisy. IEX say not having a co-lo is fairer when the move to colocation was one of the great improvements in fairness in exchange history. Latency arb potential is worse at IEX than all other US NMS exchanges and yet they'll promote themselves are your latency arb saviour. A very dark exchange.

Don't be a schmuck. Follow the money. IEX makes a fair bit of revenue from very little turnover. If you can't tell who the sucker at the table is, think harder.

So, in summary, beware of IEX's free pipe. It is not the pleasant ale you're looking for. It is directly connected to their sewer.



"These aren't the droids you're looking for." - Simon Baker


Side note: I used to work with Simon, not that he'd probably remember, as is asymmetry.


Happy trading,


--Matt.

Friday, 15 February 2019

General market stupidity - "More than $2B of Inefficiencies Found in U.S. Stock Market"

I always enjoy seeing a bit of market stupidity either in the press, paper, or policy. It reminds me
of the many stupid things I've done over the years. It is reassuring to not be alone in the world.

So imagine my joy when I hear about the free lunch theorem violation in the WSJ, "Brief Price Gaps in Stocks Cost Investors $2 Billion a Year" by our old friend Mr Cesary Podkul. Happy Valentine's Day and glory be to the great atheist in the sky.

WSJ graphic

So all those HFT firms that struggled, failed, had to sell themselves cheaply in 2016, they were just dumb? They forgot to eat their free lunch?

Mr Cesary Podkul is starting to inhabit a regular place in the silliness stakes, though he'll need to lift his game to catch my 51 stupid years of misdemeanours. I was about 15 years old when I figured out an LSE trading model with these Fourier things I'd learnt about that would grow to accumulate more money than planet Earth's money supply. The market data Mike and I hacked by bypassing the $0.10 a page update on stock prices from ViaTel in 1982 on his Amstrad PC wasn't so much the problem then. With dark parallels to that sordid tale, Mr Podkul is overly reliant on an improper interpretation in a new study for his study of that study is a case study for why studies outside your field of study should be studied further for studiously requiring study before you study. Let's meander on.

I remember Mr Podkul for the promotion of the fake WSJ IEX promotion hiding in a flawed study from, “Study Finds ‘Speed Bumps’ Help Protect Ordinary Investors” by Cezary Podkul. I meandered on that back mid last year with "IPCC wrong on climate change & SEC wrong on IEX." That was based around Dr Hu's flawed research paper published at SSRN, "Intentional Access Delays, Market Quality, and Price Discovery: Evidence from IEX Becoming an Exchange." You'd hope Mr Podkul would learn. Alas, not.

I noted back then,
"The paper highlights much of what is wrong with the academic approach to the financial industry. You take a bias and false premise and try to fit a certain outcome."
I could have said that again, but, hey, it's a blog, why pass up the opportunity to quote myself.

You'll find the two related papers the WSJ is giving dubious credit to here:
"Scaling of inefficiencies in the U.S. equity markets: Evidence from three market indices and more than 2900 securities", David Rushing Dewhurst, Colin M. Van Oort,  John H. Ring IV, Tyler J. Gray, Christopher M. Danforth, and Brian F. Tivnan, Feb 14, 2018; and,
the earlier work which shows more detail, "Fragmentation and inefficiencies in US equity markets: Evidence from the Dow 30", Brian F. Tivnan, David Rushing Dewhurst, Colin M. Van Oort, John H. Ring IV, Tyler J. Gray, Brendan F. Tivnan, Matthew T. K. Koehler, Matthew T. McMahon,1 David Slater,1 Jason Veneman,1 and Christopher M. Danforth
These are written in combination with the University of Vermont. They are copyright "The MITRE Corporation" which seems all too proud of their misdemeanours as you can see in the promotion on their web page:
(click to enlarge)
"More than $2B of Inefficiencies Found in U.S. Stock Market" they scream. That's a lot of loose change in the back of the couch. Despite the nice presentation, there is a fundamental problem here. If you think you have found something too good to be true, it is probably not true. It is not true.

As Mr Remco Lenterman tweeted:

(click to enlarge)

If only it was that simple. Amen.

Kipp Rogers, Mr @mechanicalmarkets, pointed out the wild implausibility of:
(click to enlarge)

Poor old study. Simply doesn't pass the smell test.

Market data, even direct feeds are a rearview mirror into the reality of the auction. We guess at the truth. The SIP is a delayed rendition of part of that. All of these systems have flaws and jitter as do the timing and transport systems for measurement and co-location.

If you study the timing between the ES mini and the SPDR, the simplest of arbs perhaps, you'll often figure the market data is faster than the speed of light. How is this so? Well, there are two reasons. Market data is only half the story, order processing and its associated propagation and games is another. Secondly, smart speculation with heuristics, statistics, and intelligence, artificial or otherwise, lowers the latency. Smarts improves latency with uncertainty. It is all part of your extended distribution of opportunity.

Remember, sometimes the SIP is faster than the direct feed too. I'm looking at you IEX, you dumb schmuck latency arb provider, you. Bad boy Brad. Down boy. Down.

Putting that aside, this study is pretty, petty, and shiny; but it is mainly about uneducated noise. Double entendre intended.

I'd also like to meander about the 3ms speed-bump that ICE is going to apply to silver and gold futures. That hall of mirrors argument for introducing inefficiency and poor market microstructure, on purpose as a "Hail Mary" substitute for a lack of growth, is also too stupid to waste too much of your time on. At least we financial types will have nice complexities to keep us all busy and unproductively employed.

Happy trading,

--Matt.

Saturday, 9 February 2019

"4" by Alexandre Laumonier - "Sniper In Mahwah"

What a lovely object to see in the mail!
Hauts de Seine

I'm working my way through Alexandre's book "4."

It is delightfully packaged and playfully written. The narrative style Alexandre uses draws me in. The compulsion to read more is strong in this one.

The machine translation I'm forced to use as an ignorant non-native speaker is a little inconvenient but it works well for me even if I'd be more comfortable with native English. AI is working well as my I lets me down.

I have a bit to go as it is slow going with my android/google translate from photographed images. Every extra page imaged and translated is a delight. I'm kind of enjoying the reveal of each page. It's very much a low-frequency game that helps distract from the wallabies eating the tomatoes this summer weekend in the land of Oz.

Alexandre has an extract titled "High-frequency trading networks" at Visionscarto. Here is the introduction for you,
"After the 6/5 works , Alexandre Laumonier continues his exploration of world finance with 4, a new episode of his investigations that has just appeared . This book is entirely dedicated to the history and techniques of the recent radio communication networks deployed by some traders to connect the financial markets with one another. The reason? Radio waves provide the opportunity for market data to switch from one exchange to another two times faster than optical fiber. Having a microsecond ahead of the competitor guarantees gains in what is called high frequency trading. A true ethnological and geographical survey, this immersion in the heart of European networks (mainly from London to Frankfurt) and American (linking New Jersey, Washington and Chicago) is like an epic story. 
Here is an excerpt from Chapter 3 of 4, devoted to the first ever microwave network in the United States for a trading firm."


The slight book weighs in at a little over 100 pages. I'm disappointed it is not 300 pages so I might enjoy a longer latency journey but 100 is about right for my manual imaging exercise. It is beautifully published with a seductive wide bookmark showing a silver embossed European microwave map with a grey Chicago-NY map on the flip side.

Enough for now. I have to get back to my phone imaging and translation exercise :-)

Thank you Alexandre for a weekend delight!

--Matt.

___________________

Postscript: I've finished reading "4" now. Quite the delight and highly recommended.

Lovely details and anecdotes that only someone as invested in the story as Alexandre could tell. Even those of us in the industry only have a hazy picture of experienced truths, rumour, and beer talk to go on. Alexandre fills in many details with enough fuzz remaining to keep us all curious.

Sniper touches on regulating heel sizes in trading pits, to the Chappe optical telegraph and stock scam, to William Heath on a bicycle, the "American Deer", being replaced by copper wire.

Some fun quotes, if you'll excuse the translation from French,
“Better to be first 99% of the time than second 100% of the time” 
Five years later, in Aurora, a microsecond required a $14 million investment.
Market data is now passing through the bell tower, neither seen nor known, and money from traders using McKay's network has indirectly helped renovate the church.
The presence of Jesus below the antennae of traders is common, the crosses having been, before the erection of the modern pylons, the highest structures arranged on the high points, for the same reasons as those justifying the presence of the parabolics in these places: to be visible from afar - this would make some say that, in the heights of the landscape, one religion has replaced another. 
While they were visiting the towers, the guys from Jump were asking the pilot to let the rotors spin, in case the authorities had disembarked and they had to clear off urgently. 
...the inhabitants of the neighborhood designate under the nickname of "penis." ... His erection was a technical feat...
Once on the platform, they noticed that large parabolics pointing to Frankfurt and London were already there. One of the toughest competitors of the American firm occupied the heights of the cathedral: the Canadians of Vigilant Global, who had surveyed Belgian territory with a length in advance.
rumors circulated that these officials would have asked high-frequency traders to run the restaurant in exchange for the installation of their parabolics
in addition to the blog he created to fight against the tower of Canadians, this computer scientist eventually wrote, printed and deposited at home neighbors of anti-pylon leaflets, he was cruelly bitten by a recalcitrant dog that prevented him from accessing a mailbox.
Thus, in Banana Land, the "battle of the two towers" began
The ending prose and pointed juxtaposition of two distant tribes colocating was a nice way to close a continuing saga.



Friday, 8 February 2019

Random meandering catch-up

Virtu / ITG


A few things interesting things have been happening why things have been busy/quiet here.

I see the Virtu / ITG deal has the regs done and is likely to close Q1, "Virtu Provides Update Regarding Acquisition of ITG."

This is a good out for both ITG and their customers. I doubt ITG was going to hit the plan numbers Mr Troise was espousing on each earnings call, so now they don't have to. It's good for most ITG customers as Virtu is an impressive outfit. They have much better tech than ITG which may help you'd hope. It's bad for ITG staff as their payroll has been bloated for years. It is at least 50% overstuffed with staff who may soon be stuffed. Virtu is an aggressively lean organisation, as KCG staffers will attest to, so perhaps only a quarter of staff are really needed? How deep to cut? Losing some ultrasensitive agency-only clients would then be much more palatable with a lower cost base.

The firewall between the prop biz and the broking should be safer with Virtu. In the old daze of ITG they just lied to their customers about their own trading and pretended it was an agency only business. I wrote about this a few years ago, "ITG continues to deceive customers." ITG's Frank Troise sensibly closed their prop business down, "ITG shutters prop trading business." The hits kept coming for ITG, from the $20M Project Omega mess, to $12M for the POSIT confidentiality leaks and misinformation, with the $24M for improper ADR handling in between.

ITG and their shareholders may have actually dodged a bullet. ITG prop traded for many years and lied to their customers about being agency only brokers. If the SEC had been wiser and a slightly more adept regulator they would have disgorged those tens of millions and put a juicy fine in their pocket too.

ITG was an ATS and algo pioneer. They did some exceptional work in the early daze. It aged. ITG became clunky. Their tech didn't work so well, but many of the buy-side remained. Sticky clients. Sub-standard tech and ridden by scandals. People were all too attached to their Triton terminals.

Virtu is buying those sticky clients. Some clients may run away, but they shouldn't. Virtu is sharper and should provide a better future. Many possible acquirers have run the ruler over ITG for many years. I've talked to a few. The danger of catching a falling knife has kept many away. The deal should work for Virtu as it can aggressively pare costs (aka staff), they seem a little smarter, certainly have much better tech to integrate, and thus Virtu's outcomes should be a better business. It is ironic that the buy-side seemed to trust the dodgy ITG and be wary of the Virtu HFT machine. It will be a leap for the buy-side to trust an HFT. The truth of the matter is different. Virtu's reputation is strong and ITG's was kind of sucky until Mr Troise started cleaning shop. New customers may actually find a reason to call on the embedded ITG again. That would be a change to just sticking around.

I suspect a win for Virtu.

DRW


As a bookend to the CFTC critical meandering, "CFTC vs DRW: regulator's ego explodes" I should note the conclusion. Mr Don Wilson scored a big win when US District Judge Sullivan wrote,
"It is not illegal to be smarter than your counterparties in a swap transaction."
Crain's covered it here, "DRW's Wilson scores big win in market-manipulation case." It took an inordinate time to conclude, as is the want of courts. The judge had a new job confirmed by the Senate so he had to clear the decks of what he could.

The CFTC does so much good work and then let some legal eagle's ego run away and allow a train wreck such as this. Even with such an obvious case, DRW had to risk much to litigate this to the mat. Good on them for not folding and holding the CFTC to account. There was a change in legal personnel at the CFTC. Heads should have rolled and they did. I doubt Mr Wilson got the apology he deserved.


Volatility is your friend


Around December there was much talk of computers exacerbating volatility on the downside. Very little talk about algorithms with upside volatility. That is the press norm.

What many people miss is that better information and models should lead to a more aggressive view of valuation from a bottom-up perspective. Think of the sum of discounted cash flows point of view. Share price should be very sensitive to initial conditions, i.e. today, whether that be earnings or new insight, macro or micro. We had the crazy volatility doldrums where any dope could make money selling vol until February 2018. Then we start blaming algorithms and computers.

The real story may just be that we are getting better at working out value. Thus prices incorporate information more efficiently. Sharper price changes result. That is, efficiency may just be volatility. Welcome to the new world.

I'm not convinced but it is interesting. I suspect noisy crap is just noisy crap. Signal is hard to diagnose.


Arista bought Metamako


I had a few beers with Charles and Scott after they baled with the sale back in September last year. Scott had just bought a new toy.
Scott Newham (left) and Charles "First time" Thomas (right)
Scott's new toy with two wheels.

Dave Snowdon and the rest of the team stayed on to play with their new Arista toys and teammates. This is significant in the HFT world as most profitable firms use their equipment. The newly named Arista 7130 is the go-to device for layer one.

It is a good deal for Arista. They get back to their financial service roots. I remember meeting Andy Bechtolsheim at the Roosevelt in NY many moons ago when Arista had a simple trestle table and a few pamphlets. I felt giddy shaking his hand. Geek I am. The crude marketing setup belied the $100M Andy personally sunk into the venture. It was the fastest switch at the time with Fulcrum's asynchronous wonder chip. Arista did well. They started by displacing Cisco with fast Fulcrum based switches that only did layer two. It was a while before Arista got to layer 3. Now they are getting into big iron routing which looks more like a huge switch in the 21st century.

Metamako displaced some Arista gear from financial services with clever layer one and some faster layer two, borrowing from the same Arista playbook.  The same plot worked well with new actors.

The Metamako deal opens some doors for Arista. Metamako's plans have a great enabler in Arista with their super-sized resources. The tango looks sweet. I suspect a happy ending. Well done all.


Personal note


Not the best year last year. Not sure where it went. Marriage nailed shut. Aged mum has lurched from stroke, recovery, cancer, recovery, to death by dementia over the last couple of years. Not a great route but an all too familiar sadness for so many. Dementia became the largest killer of women in Australia in 2018. We should do something about that.

An old high school friend, Mardi Dungey, who was a Professor of Economics for a while at Cambridge before coming home to a Professorship at the University of Tasmania got a shock diagnosis late last year and a sad memorial last Friday. She was one of the good ones. Fiftyish is way too young. Life is fragile.

My old father survived his carotid artery surgery three weeks ago and recovers slowly. Then the fires came to the Huon Valley just as the flood renovations were completing.
A firestorm brewing beyond the front yard.
Lucky with winds at the end of the day.
Death threats make for a welcome sunrise

Relying on high latency canaries for fire signalling
Roads opened up again today. An inch of rain yesterday may have had something to do with that. A few houses were lost but the impressive fire-fighters largely won their battle.
Smokey river. The old man has steroids for an asthmatic cough now.
I'm waiting for the 'roid rage to kick in.
A bit less smoke
All clear again
The canaries were stood down from alert duty today. Sadly, one juvenile was lost.
Though it was nastier when the tiger snake took three young ones a couple of months
ago. Stupid tiger snake died a death of gluttony as he couldn't escape the cage.
He met a garden implement.

Normality instead of this shitty leptokurtosis would be nice for five minutes. I guess things are different here: our black swans have white wing tips:

What distribution can be derived from black swans with white wing tips?


Happy trading and let's all have a good 2019,


--Matt.

IEX 2019 - *sigh*

IEX is a nasty business.

The expensive dark exchange, which is still not a restaurant, continues to push false rhetoric and mislead investors. Nothing new there. This time it is spurious indignation surrounding market data.

IEX is a small exchange with such poor execution it is surprising brokers have not yet been prosecuted by the SEC for a lack of best execution for routing orders to IEX. When will the SEC honour its regulatory obligations? A fair-minded person would see the SEC needs to prosecute for false and misleading statements or a lack of best execution.

I've covered much of this silliness in pieces such as "IEX's Data Core delusions."

Nasdaq added to this noble cause by publishing a piece worth reading on IEX's poor execution quality here: "Three Charts Dispel the “Price Improvement” Myth."

IEX has lied to the market and its users in many ways. I've meandered about this many times before but then IEX pipes up with a fake narrative around market data which should annoy any market professional. It has annoyed me enough to call out their continued fakery yet again.

In a very old meander about speed bumps, "Speed-bump 101", IEX's speed bump weakness is covered. Not having a fair co-lo is normally enough to give the well endowed an advantage but that is not enough for IEX. IEX enables the worst exchange latency arbitrage play within the US NMS by delaying market data to their customers despite enabling immediate SIP updates. Oh dear, IEX. The latency arb enabler truth does not align well with IEX marketing.

IEX doesn't allow you to trade fairly when the market is changing. You'll be prevented from trading or penalised with a high transaction price. That high price is more often charged for no reason as their crumbling quote indicator is often wrong. This mechanism also makes their fee unauditable as not only do you not know what you may be charged apriori, the lack of visible determinism means you may not even calculate your fees correctly post hoc. A sad state of affairs.

IEX offers rebates, which they call discounts, if you qualify. Yet their rhetoric on others' rebates was misguided and insulting as they sought to carve out a marketing advantage without fair reason.

IEX market data is pretty useless. Not only is it delayed, but the uselessness is extended by often avoiding a proper representation of NBBO due to its lack of market best quotations. The market data is free which is good. Perhaps you get what you pay for here. It is certainly useful if you want to get an approximate idea of where the market is, but you have a smartphone for that too.

BATS market data used to be free. This was a very good deal. Eventually BATS started charging for market data. Dave Cummings signalled disappointment that BATS did this. It was a shame.

The inelasticity of demand has caused exchanges to pump up market data fees at a rate which is far from reasonable. There has been much noise about this and finally the cries SIFMA et al cut through and the SEC has pushed back on the exchanges unjustified fee rises. This is how it should be.  That inelasticity in demand, you don't have to trade everywhere but you need market data from everywhere, insists that market data prices need to be regulated.

So IEX jumps on this bandwagon and steps too far. They write a hit piece on how inexpensive their two-bit exchange delivers somewhat useless market data and falsely compares that to real exchanges. There is much that is disgusting in that piece but also a little bit of truth, however slender. It is true the exchanges have bled that demand inelasticity too far and they need to be regulated, but stupid numbers from IEX do not make them the white knight that will save us. IEX remains not only a terrible exchange to trade at but an exchange that continues to have a mendacity problem. IEX  cannot help itself but mislead further.

*sigh*

Anyhow, now I've got that out of my system. Let's meander through where IEX currently sits in the market space.

First of all, the Nasdaq v IEX patent lawsuit rolls on. The settlement conference failed as expected. Judge Martinotti granted IEX's dismissal of the induced infringement claims but denied IEX other dismissal requests in an opinion filed on January 4th. Nasdaq filed a 67 page amended complaint on January 25th which continues to demand a jury trial.

If you squint it looks like IEX market share continues to grow, albeit slowly, as it continues to strive for a monthly average of 3% but continues to fall short:
(click to enlarge)

To IEX's credit, there have been a couple of days when less than 70% of its handled volume not lit. Not exactly something to be proud of, but some progress. Nevertheless, the sad exchange is still averaging about 25% of the handled volume being lit as you can see below. January slipped back to 23.6% of handled share volume being lit. It is an expensive dark place indeed.

Happy trading,

--Matt.


(click to enlarge)