Thursday, 10 December 2015

There you go: speech recognition solved

Deep Speech 2 architecture
(click to enlarge)
Just "Wow!" results from Andrew Ng and the team at Baidu. The first significant publishing of human competitive speech recognition performance is a real breakthrough, even if other non-public results may be lurking.

The paper is, "Deep Speech 2: End-to-End Speech Recognition in English and Mandarin"  by Baidu Research – Silicon Valley AI Lab∗, Andrew Ng, etal, pushed to arXIV recently.

Google with Geoffrey Hinton, Andrew Ng, and others started much of the ball rolling with outstanding ImageNet results that are now human competitive. The father of getting Fukushima's Neocognitron from 1980 to evolve to learning as a convolutional neural net, Yann LeCun, has his team at Facebook beating human level performance on facial recognition. So, automatic speech recognition (ASR) getting to human level performance is not unexpected, but it is still a major achievement with enormous ramifications for the way we will work and interact within the world.

From the introduction,
"The Deep Speech 2 ASR pipeline approaches or exceeds the accuracy of Amazon Mechanical Turk human workers on several benchmarks, works in multiple languages with little modification, and is deployable in a production setting. It thus represents a significant step towards a single ASR system that addresses the entire range of speech recognition contexts handled by humans."

A nice grab from the paper showing human competitive performance.
(click to enlarge)
From the conclusion,
"End-to-end deep learning presents the exciting opportunity to improve speech recognition systems continually with increases in data and computation. Indeed, our results show that, compared to the previous incarnation, Deep Speech has significantly closed the gap in transcription performance with human workers by leveraging more data and larger models. Further, since the approach is highly generic, we’ve shown that it can quickly be applied to new languages. Creating high-performing recognizers for two very different languages, English and Mandarin, required essentially no expert knowledge of the languages.
... 
We believe these techniques will continue to scale, and thus conclude that the vision of a single speech system that outperforms humans in most scenarios is imminently achievable."
Much still to be done, but that is just work. Exciting times.

--Matt.


Low cost e-commerce not possible in Australia

I wished to try out a PAM8403 audio amplifier IC. It's a modest 3W class-D audio amp that is very common. So onto E-Bay and Alibaba to have a look for an inexpensive module. The picture here shows the most common form factor I see. I ordered one from Shenzhen and hope it turns up like all the other little $0.20 to $2.00 items that magically materialise, usually at 6AM, by the front door.

One of the common PAM8403 modules
This module cost $AUD 0.48, including international postage, to buy and get delivered. You can do better per item if you need ten. My budget, aka wife's credit card, doesn't stretch that far.

Whilst the module may only cost a few cents to build, the total cost, module and postage, is below the rate for posting an empty letter domestically in Australia.

It is a remarkable logistics story as this is not unusual. If you're prepared to wait the 2-8 weeks for such little things, you're better off getting them from Shanghai, Shenzhen, or Hong Kong. They almost always turn up if you stick to reputable (i.e. highly rated, thousands sold) e-commerce front ends.

China Post is the real story here. They provide the smart logistics and bulk bundling that empower this machine. There is certainly a hint someone must be getting exploited in this distribution system to make it work like this, but who knows.

You can make a clever product, completely automate the build, and have competitive input costs but establishing distribution is always a problem. Direct is best for avoiding rent on distribution, but that is not always possible or wise depending on the product. What is undoubtedly true though, the spiderly low cost logistics of a China Post give SME's an unprecedented ability to start-up and grow, as long as you build and distribute from China or HK.

--Matt.

______________

PS: It appears much of the credit, including for enabling ePacket in China/HK, goes to the Universal Postal Union which is now part of the United Nations. International mail was originally free when it arrived at the destination, but now there is a nett bulk weight differential paid, of approximately $1 per kg. The Straight Dope of 1990 vintage has a decent short explanation. It seems a clever hack for a sovereign nation to assist exporters take advantage of the UPU agreements, if only Australia Post could do the same...

Sunday, 8 November 2015

ITG - good recovery

The earnings call, capital management, and spin from ITG has been working well. The share price settled on Friday at $20.60 which is 63% up from the recent low.

Google finance link
The spin has been a little over the top on the so-called recovery, but that represents good work from ITG as it is their job to do exactly that.

POSIT numbers from the Finra stats, two weeks delayed, have not really recovered. Quite a bit of their loss of market share has been masked by dramatically higher market volumes across the industry assisting the revenue side of the equation. ITG mentioned on the call that the second half of October saw improvement which is not in the Finra stats yet.

We don't know if they have stopped their prop trading of US/CAD interlisted stocks as they simply stopped mentioning it, or breaking it out, in their reporting a couple of quarters ago. You'd be hoping there was no-mingling of their prop trading with their new interlisted stock service they've recently started marketing ;-)

The energy research sale was a good result. Obviously earnings going forward will be affected but they need the cash. As was pointed out on the call, the other research would be a neg profit line so it remains to be seen how that will be managed.

There remains uncertainty from the business with respect to legals with two class actions and various staff/ex-staff legal issues. This was the quote from the call,
"In connection with the SEC settlement, two shareholder lawsuits have been filed against ITG and certain current and former executives seeking unspecified damages. Separately, ITG's former CEO has filed a demand for arbitration claiming that his termination for cause represents a breach of his employment contract. This arbitration demand seeks an award of damages in equity totally approximately $8 million plus an additional $5 million in punitive damages. The Company intends to defend itself vigorously in all of these matters and we have not set aside any reserves in connection with these matters."
That Gasser remains on the board whilst suing the company is an interesting legal conundrum.

There are a lot of good staff at ITG so the recovery is good news for those deserving staff. However, how the board members responsible for authorising the SEC prop trading fiasco remain raises questions. It seems it is OK to mislead and defraud your customers as the obvious apathy in ITG's case means they don't mind too much.

Tuesday, 27 October 2015

More lies and misdirection from Exablaze & Zomojo

Exablaze / Zomojo made an improper statement regarding the legal status of the dispute with myself.

They have publicly pronounced, "This now brings the matter to a close."

This is wrong. Justice Beach was a dissenter with regard to accessorial liability, writing after Besanko & Gilmour JJ on the the Hurd Senior contempt matter, and this makes an obvious appeal to the High Court of Australia warranted.
Justice Gilmour - doing good work in the community

As anyone can see from the notes about apprehended bias, here, there are matters that are not fully resolved from incorrect interpretations, not just limited to paragraph 73, in the main FCAFC judgment. Not withstanding this, there is also a High Court worthy challenge for a clarification concerning apprehended bias as a matter of national significance with respect to the interpretation of the Australian Constitution based on the Judicial Employment Paradox I have previously discussed.

Further to this, at least one of the judges was impaired during the hearing which is yet to be challenged.

We may have many months or perhaps more than a year to go on this matter to get it properly resolved.

--Matt.
_________________

And now for something uplifting which my middle daughter inspired me with when we were both suffering a little (we often play it in the car now):
...
I heard you don't have to wear a suit to argue the truth
And so
Who are you to tell me how to live my life
Cause I won't give this up
These are my shoes
My view
My cue
To say I do give a fuck



Sunday, 25 October 2015

Corruption, Zomojo, Greg Robinson, Gordon J, and the Federal Court

The tragedy of the commoner.

My appeals raised last year and heard on March 3 & 4 had their judgments handed down last week. The judgments were harsh, beyond reasonable belief, and riddled with errors, inconsistencies, and invective.

My first reaction was to be pretty pissed off. Now a few days have passed. The blood simmers rather than boils. I find myself reflecting on how it could go so wrong. It may have been a reasonable result for the JJs to say, “as a self-represented litigant you're an incompetent fool” and dismiss the appeal. I could have respected that.

They did not. They piled on unreasonable and malicious reasoning with intemperate language and flawed logic.

The first question that comes to mind is that is the system a giant conspiracy and out to get me?

No.

The second question to ask is if the JJs are bad judges, full of malevolence, and corruption?

No.

These judges are good people.

Gilmour J was distracted, inattentive, and somewhat regard free. Beach J was impressive but clearly prejudiced. Besanko J simply did not care enough to seek the truth. Nevertheless I've seen the work they do and these are good people who work hard for what they see as justice.

The question remains that if these are good people that made such a malicious judgment, where did their motivation come from?

It's pretty simple on reflection.

Honourable Justice Michelle Gordon's original bias and toxicity has simply infected everything that has subsequently flowed. The JJ's simply believed in Gordon's findings and felt they were doing society and justice a favour by burying me. Hurd the toxic, begone.

The conclusion is pretty simple to reach as you can see in paragraph 73 of their judgment they had no basic understanding of the case. They simply didn't bother. The case seemed to be deemed a distraction from a self-represented litigant who was calling out a biased judge, their friend and colleague. The matter was heard at the end of their appeal circuit and was simply a puss filled boil that was preventing the closure of their junket and further time at the bar. They lanced the boil.

The judges certainly believe they have done the right thing with their inattentive misunderstanding. It is a shame that the legal system works, or fails, in this way.

Davies J decision in 663/2013 was much worse and more perverted but more understandable as Her Honour was:
  1. being fed by corruption within due to Registrar Burns being a friend of Robinson;
  2. worked with Gordon J in the same small department in University Melbourne Law; and, 
  3. no doubt “owed” Gordon J somewhat due to not only the friendship but the recent appointment to the Federal Court and assimilation.
The JJ's decision is simply on the same progression. No understanding of the case and a desire to project a protective shield over criticism of the court and the influential Hayne & Gordon power couple.

The whole saga is representative of the corruption in society that is prevalent when institutions mature and the politics of appointment and favour move the dial from the meritorious to the connected. Australia has long been far from the egalitarian ideal we often fool ourselves into believing. This case shows just how that corruption has become endemic to our society. Not the graft, $ in hand, kind of corruption, but the connected, influence, self reinforcing favour, and status kind of corruption. This case has to be the poster child for exactly that decay. The country is unsafe.

There can be no expectation of public confidence in the legal system.

Australia is dead.

Fiat justitia,

--Matt.
_________________

Some details on the apprehended bias I failed to fight against:

Even though Zeptonics and Zomojo are based in the Sydney, the matter is raised in Melbourne.

Robinson has prior boasted of the influence he can bring to bear in Melbourne but there seems little to worry about as the case is largely without merit.

Honourable Justice Michelle Gordon has the case.

Her Honour's husband Honourable Justice Ken Hayne is a High Court of Australia jurist. Hayne is a good mate with the Governor of Victoria, Hon Alex Chernov, both being ex Vic Supreme jurists. Hayne swore in Chernov as the Administrator of the Commonwealth of Australia, for example.

Bill Burdett, Zomojo beneficiary or shareholder, friend of Robinson, works in an excellent charity, Friends of Asha, alongside Chernov.

Greg Robinson and Bill are ex-ITG folk. Bill was on the NY board as a Director, Asia-Pac chairman, and Greg ran ITG Asia-Pac as CEO. I met them in 2005 when I was looking for a job and ITG was considering using me to set-up a prop trading biz in Sydney to complement their existing prop trading in the US and Canada. Bill and Greg decided to take the opportunity for themselves instead of ITG which is how Zomojo was born with me starting up with their capital investment.

Still, these connections the Zomojo guys have should be not that much of a concern, should they? Justice should still work?

The judge Michelle Gordon, works at Melbourne University. Greg Robinson's wife, Patricia Desmond works at Melbourne University as a Prof and runs a medical department there. That's a bit more of a worry. Two senior and powerful University women.

Greg and his wife have donated at least $300,000 to the university and mix it with Gordon and her ilk at the university, including Michael Robinson AO, Gordon's former boss at Allens. Then again, maybe they don't even know each other?

Greg's wife has done work for Michael Robinson AO at his health concern, Epsworth Medical Foundation, he chairs. Patricia has invited him to gala dinners. Perhaps they do know each other after all.

Greg Robinson provides an affidavit that says he had lunch with Michael Robinson at the men's only Melbourne Club.

It turns out Hayne J has a bunch of golf buddies, from that den of paedophilia known as Scotch College, that he plays weekly golf with and some of them, such as Robyn Syme, are mutual friends of Zomojo's legal team (including Janet Whiting), Greg Robinson, or Bill Burdett.

The Melbourne club is a tight knit group and all these people know each other and hangout in the same circles. National Gallery of Victoria. National Australia Day Council. They bump around and feed off each other. That is not that much in itself but the donations of at least $300,000 to the university seem to raise a bit of a worry as, in effect, Robinson has been giving money to the judge's employer. This seems a conflict too far.

Now the original case was run in October 2012. In 2012 a lass by the name of Kate Dillion (nee Desmond), a close relative of Greg Robinson and his wife Patricia Desmond, was awarded the Melbourne Law School Masters Award – Davies Collison Dave Prize for 2012. So, Gordon J's school gave Robinson's relative, where he donated lots of money, a prize. No influence?

It should seem to you less and less likely that a fair trial could be in the offing.

Now the Zomojo barristers are Dr Matt Collins SC and Dr Michelle Sharpe. They seem good people whereas Janet Whiting and Matthew Critchley, Zomojo's solicitors, are outright lying scum.

Hold it a sec. Matt Collins works at the University of Melbourne Law School. Michelle Sharpe has worked there too. Goodness, how far can the conflicts lead us? Really, should it be allowed that it is appropriate for both opposing counsel to work with judge? What the ????

No wonder Zomojo's counsel is so favoured in trial by the judge despite their lies. They are colleagues and friends of the judge!

We get an answer to an interlocutory application, including a Robinson affidavit, before it has even been stamped and received by us. That is strange as it has certainly not been served. We complain. We get another copy of the interlocutory application, this time with a new date stamp on it back dated to the prior week. Hmmmm. Something fishy is going on in the Victoria Registry. Smells of a little corruption.

In VID 663/2013, the Zomojo liquidation matter, Registrar Burns is actingy harshly and weirdly against us prior to trial. Burns declares that Greg Robinson is a friend on the Friday before trial. He does not recuse himself and continues to act against us. This perhaps explains why information and documents have been leaking to the other side all this time. Moles, leaking, favours, corruption. What the ???? Seems a little ridiculous. Why is this case so special?

Besanko, Gilmour, and Gilmour JJ's have ruled this is perfectly OK. Even though, Gordon J acted harshly against us, ignored evidence, granted mistake riddled subpoenas to Zomojo whilst being a pedant with ours, changed the procedures on the fly in favour of Robinson, hardly granted us any objections and granted almost all of theirs, even ignoring evidence that didn't suit Her Honour's prejudice along with many "inconvenient" legal submissions and precedents… there was no possibility of impartiality.

Some of these facts only became known this year, but then again, we should have known we were in trouble when Gordon J made snide remarks about New South Welshmen in court. It was unlikely to end well, and it did not.

The system is corrupt.


______________

Sometimes they try to break me down
(hey-ooh)
Other times I feel safe and sound
(hey-ooh)
And I don't know why I'm weightless now
But I'm guessing it's because we're running on air


I was told that life is as easy as can be
But apparently the tragedy is freedom isn't free

Thursday, 22 October 2015

Exablaze & Zomojo's Greg Robinson identified as a liar by the Federal Court

Some small solace in my continuing battle was that Dr Greg Robinson, former ITG Asia-Pacific CEO, current Chairman of Exablaze and Zomojo, was found to be a liar by the Federal Court of
Australia.
Exablaze's Greg Robinson?

Small beer in the greater scheme of things perhaps, but it is the first acknowledgement by the court of the fact that Greg Robinson is an unsavoury character.

From paragraph 87,
"What this sequence discloses is that there was a genuine credibility issue that was raised with Dr Robinson on this matter.
This is legal speak for unctuous lying scum. Strangely, the court concludes that the issues of credit with Robinson are of no consequence. Yep. Bizzare. Just as court also concludes that the fact that Mr Matthew Critchley of Corrs Chambers Westgarth lied to the court is of no consequence.

Despite such activity being a criminal offence, the lesson clearly is that it is best to lie as often and advantageously as possible to the court as there are no consequences.
Caveat emptor,
--Matt.
PS: I've previously written about part of this here: Corrupt lawyers - why you should seek them out.

Hurd hit with a suppression order from Federal Court

Apparently the evidence given in my sworn affidavit to the Federal Court of Australia's appeal court is so controversial it had to be placed under a suppression order preventing its publication so that the world would not stop spinning.

Cronyism, bias, corruption, and cover-up is alive and well in the Federal Court of Australia.

The details of the order made by Honourable Justice Beach read:
  1. Pursuant to s 37AF(1) of the Federal Court of Australia Act 1976 (Cth), subject to further order the affidavit of Matthew Hurd sworn on 5 January 2015 together with the annexures be prohibited from publication.
  2. The ground upon which order 1 is made is that contained in s 37AG(1)(a) of the Act.
As far as I know, there is no order preventing the publication of this non-publication order.

--Matt.
___________________________

FEDERAL COURT OF AUSTRALIA ACT 1976 - SECT 37AG

Grounds for making an order
             (1)  The Court may make a suppression order or non-publication order on one or more of the following grounds:
                     (a)  the order is necessary to prevent prejudice to the proper administration of justice;
                     (b)  the order is necessary to prevent prejudice to the interests of the Commonwealth or a State or Territory in relation to national or international security;
                     (c)  the order is necessary to protect the safety of any person;
                     (d)  the order is necessary to avoid causing undue distress or embarrassment to a party to or witness in a criminal proceeding involving an offence of a sexual nature (including an act of indecency).
             (2)  A suppression order or non-publication order must specify the ground or grounds on which the order is made.

Re: Federal Court of Australia Corruption


Appeal judgment out from the March hearing. Lost badly as expected against any residual flicker of a hope for justice. Example:
73. The chronology of events discloses a number of matters. First, whatever may have been the position before trial or during the hearing concerning how the cross-claim was to be dealt with, as matters developed given her Honour’s principal findings, both parties considered that an order for dismissal of the cross-claim was appropriate given her Honour’s principal findings. No party submitted otherwise. Indeed the respondents below proffered such an order to her Honour. Second, as the chronology discloses, the respondents below were given every opportunity to submit to her Honour that her principal findings did not justify a dismissal of the cross-claim or should be revisited in the light of the orders made prior to trial on the question of how the cross-claim should be dealt with. No such submission was ever made, and even though the respondents below were represented by both counsel and a major law firm. Hurd’s assertion that the respondents below had no choice but to agree to such an order is not credible; the respondents had an opportunity to object but chose not to do so.
WTF? You would think only a triumvirate of imbeciles that failed moron school could write that. Besanko, Gilmour and Beach JJ are not imbeciles. This does not make sense. Unless, malice is the motive.
The wagons circled around corruption and bias at the Federal Court of Australia.
Justice begone.
--Matt.
__________
PS: Background. Judge found that the cross claim to be dismissed. Ordered both sides to bring orders to Her Honour such that this would be effected. You couldn't refuse as you'd be in contempt. Then appeal court finds that you're obviously guilty because you provided such an order. It's like being forced to write your sentence even though you're fighting to prove your innocence and then being found guilty because you wrote your sentence. Bizarre example of corruption and bias in the Federal Court of Australia.

Lesson (1): if you have a choice don't run cases or mediation in Australia as its legal system is expensive, untimely, and corrupt.

Lesson (2): if you have in interest in justice you should either leave the country and live somewhere else; or, take advantage of the corruption and become a crime lord.

Thursday, 1 October 2015

Simple, quick Reg NMS thoughts for the day

There are too many non-public venues and too many complexities around various venues, public or not.

Make an ATS a time limited "pre-exchange" license. Perhaps one or three years. The venue should become a public, open exchange or pack-up and go home. Principle: markets should be fair and open.
Depiction of traders under the buttonwood tree outside 68 Wall Street
(Source: Wikipedia / Library of Congress)

Block venues, currently called an ATS, should be separately licensed and have a minimum size or value threshold for an order. $1M? Perhaps they should just go away or be limited to overnight gatherers that then cross the next day, or the next week, at a fair price such as VWAP. If you don't like it, use algos to trade on public exchanges.

Payment for order flow (PFOF) used to make sense in a less mature world but now it is a perversion preventing fair and open markets, especially for retail. Most retail orders in the US never get to an exchange. This is wrong. PFOF should only be allowed for small brokers as a dip of the hat to small brokers' cost structures. Big brokers should be judged on how they serve their customers. Principle: trades should not be unduly siphoned off from public and fair market places.

Pricing for all exchanges should be one volume tier. Order type and passivity differentiation is OK if not linked to volume. If an exchange needs to consider economies of scale, do it with connection or membership fees as a fixed monthly or annual cost. Principle: fair and simpler markets.

All order types should be replicable from a connected co-location with simple order types, such as limit or ioc, with no disadvantage. Complex order types have unintended consequences and impose complexity on the client. I could imagine an SEC approved list of order type families being limited to perhaps five or ten. If the SEC approves a new family, then one gets kicked out. All other order types would have to be composed from these and be replicable by a client of the exchange with no disadvantage. Participants, including retail, should not be disadvantaged by complex order types. Principle: fair and simple markets.

If delays are used in a matching process, all information should be similarly delayed. That is, there should be no preferential information within the matching boundary, even for an exchange. There shouldn't be complex time based reasoning about time differentials in a matching process. Time differentials are akin to "knowing the future" for a decision and distort reality. Principle: need to simplify and keep markets fair.

Brokers need to do 15c3. A broker needs to understand their clients' credit. Exchanges, as so far as margin or capital requirements for a broker, should do so as well. Perhaps allow third party risk evaluators (somewhat replacing prime brokers)?

Buttonwood Agreement: The first version of Reg NMS?
(Source) (Click to enlarge)
Internalisers, as an ATS, should be a public exchange or shut down. Private parties that wish to exchange ownership should do it transparently in a public forum if they make more than some number of very low number transactions per day. You should allow private exchange of ownership but private trading, whether or not it is under a buttonwood tree, should be limited. Principle: fair, public, and open.

Not sure what to do about routing... perhaps it should be the responsibility of public exchanges or a system closer to that in Canada.

A quick 15 minute dump (or 30 minutes with pictures) until I change my mind tomorrow,

--Matt.

Tuesday, 15 September 2015

UBS and IEX gain - ITG lose - FINRA ATS Tier 1 Weekly Stats

The new weekly FINRA ATS Tier 1 Stock volume report came out yesterday for the week of Aug 24. It was a big week for volume in the markets. The total number of shares trading in the pools increased 37% from 3.7B the week prior to 5.1B for the week of 24 Aug.

As you can see from the chart below, UBS gained share to become the #1 ATS for Tier 1. IEX's gain in share was also notable though its ranking remained steady at #3.

The ITG POSIT slump continued. You can see their share in the chart below as the big orange line. In a week of improved volumes,  ITG's volume actually decreased. ITG's share dropped to 1.62% to be 67.1% below the recent minor peak of 4.92% share reported 20th July. ITG POSIT would now need to better than triple its market share to regain the ground it has lost since 20 July which is a similar story to that recently reported in ITG's August report regarding ITG POSIT Alert, their large block platform.

ITG POSIT market share summary

Date Volume Tier 1 ATS %
Total Volume Tier 1 ATS
24 Aug 82,129,300 1.62%
5,078,863,139
17 Aug 82,526,400 2.23%
3,708,216,728
10 Aug 78,325,400 2.41%
3,247,886,460
3 Aug 96,944,300 2.62%
3,693,676,062
27 Jul 144,365,800 3.88%
3,721,476,946
20 Jul 175,855,300 4.92%
3,575,618,895
13 Jul 142,353,700 4.57%
3,115,440,140
6 Jul 167,849,900 4.84%
3,467,172,754
29 Jun 127,840,200 4.76%
2,687,323,287


(Click to enlarge)
ATS data is provided via www.FINRA.org/ATS and is copyrighted by FINRA 2015


Thursday, 10 September 2015

ITG - POSIT Alert down 68% for Aug - tripling would not recover to normal

ITG released their usual monthly update on trading volumes via an SEC Edgar 8K form today.

The filing fills in a bit more of the picture of declining volumes that has been discernible from the FINRA ATS volumes. Let's meander through.

Firstly, we need to consider the market environment.

The market was busy in August, in case you hadn't noticed. The consolidated market was around 7.92 billion shares average daily volume (ADV) for August 2015, which from August 2014 at 5.24B ADV, represents an increase of 51% over the same month from the prior year. August 2015's consolidated volume represented an increase of 17.5% from July 2015's 6.74B ADV.

July was a busy month for ITG with 16.2M ADV for the block oriented POSIT Alert and 101.0M ADV for regular POSIT.

All things being equal, ITG should have seen increased volumes in August given the market state, except for that little old SEC report and settlement. That takes us to ITG's 8K filing:

ITG U.S. Trading Activity




# of
Trade
Days

Total U.S.
Volume

Average U.S.
Daily Volume

Average
POSIT
Daily
Volume

Average
POSIT
Trade
Size

Average
POSIT
Alert Daily
Volume

POSIT
Alert
Average
Trade Size

POSIT Alert
Avg. Trade
Size Ex-
Algos*



















August 2015

21

2,806,911,287

133,662,442

55,991,590

241

5,095,790

15,595

29,087



















Year-to-Date:

167

30,318,518,530

181,548,014

89,134,210

261

14,671,070

15,586

33,926


*Excluding shares crossed through POSIT Alert from ITG algorithms

The year to date figures include August, so if we back those out we get:


# of Trade Days Average POSIT daily volume

Average POSIT Alert daily volume
August 2015
21 55,991,590 -40.37%
5,095,790 -68.25%








Year-to-July:
146 93,901,299

16,048,336

Remarkably if POSIT Alert volume now tripled, POSIT Alert will not have recovered to the prior levels. Quite some remediation is required.

The scandalous SEC settlement came out on Aug 12. So, although a prospective settlement was announced late July, the market's full understanding did not kick in till half way through August. Offsetting this overstatement of "current" or immediate volume embedded in the monthly statistics is the fact that the ITG DC had shut-down temporarily affecting volume.

This is just for the US. The 8K reports the following with regard ex-US business,
"The average daily commissions in August 2015 in ITG’s Canadian, European and Asia Pacific businesses was down a combined 35% compared to the second quarter of 2015."
This comment,
"During August 2015 there was an increase in the percentage of trading activity from sell-side clients as compared to the level in the second quarter of 2015, significantly reducing the overall average revenue per share."
indicates that besides an exaggerated negative effect on revenue, the the buy-side has been the larger portion of the volume disappearance. You have to wonder why the buy-side trades with ITG at all, especially knowing they are paying more than the sell-side.

The decrease in volume is worse than it seems as market volumes have increased significantly.

The ITG share price finally crashed through the 52-week low today settling at $14.21 which was -7.85% for the day. The Board members who were sitting prior to 2011 have still not resigned. Why is Gasser still listed at being a director on the board? Does ITG sill prop trade? Did ITG mislead the US Senate in testimony? The class action law suits continue brewing.

It's not an easy time for the many good staff at ITG let down by their board and a few senior managers.

Tuesday, 8 September 2015

ITG - POSIT market share decline continues

The market share of ITG's POSIT dark pool slipped a place in the rankings to fifteenth according to the FINRA data released today, being for the week of 17 August 2015. POSIT had a small decline in market share to 2.23% of all ATS Tier 1 shares by volume which is a loss of approximately 55% of market share since 20 July 2015.

The following board members have not resigned from ITG's board; they should:

  • Maureen O'Hara (Jan 2003 -)
  • Minder Cheng (Nov 2010 -)
  • Christopher V Dodds (Jun 2008 -)
  • Robert C Gasser (Oct 2006 -) 
  • Timothy L Jones (Mar 2005 -) 
  • Steven S Wood (Feb 2010 -)


ITG POSIT statistics for Tier 1 US Stocks - released by FINRA 8-Sep-2015


Date Volume Tier 1 ATS %
17 Aug 82,526,400 2.23%
10 Aug 78,325,400 2.41%
3 Aug 96,944,300 2.62%
27 Jul 144,365,800 3.88%
20 Jul 175,855,300 4.92%
13 Jul 142,353,700 4.57%
6 Jul 167,849,900 4.84%
29 Jun 127,840,200 4.76%


ATS data is provided via www.FINRA.org/ATS and is copyrighted by FINRA 2015

Tuesday, 1 September 2015

ITG Posit - market share now 50% below peak

The latest Finra Tier 1 ATS Stock data was released 31 August. It shows the ITG Posit share slide continuing after the scandal announcement.

However, the slide from the previous week is small and within the bounds of regular noise in market share.

The notable statistic is that the 10-August share of 2.41% is over 50% below the 20-July recent pre-scandal disclosure peak. It is 50.5% of the four weeks of pre-scandal market share. You'd expect that the recent data centre problems and temporary shut-down of Posit will not help the statistics improve in a months time.

ITG POSIT Tier 1 Statistics - FINRA release 31 August 2015


Date Volume Tier 1 ATS %
10 Aug 78,325,400 2.41%
3 Aug 96,944,300 2.62%
27 Jul 144,365,800 3.88%
20 Jul 175,855,300 4.92%
13 Jul 142,353,700 4.57%
6 Jul 167,849,900 4.84%
29 Jun 127,840,200 4.76%

(Click the graphic above to enlarge)
ATS data is provided via www.FINRA.org/ATS and is copyrighted by FINRA 2015

Friday, 28 August 2015

Trade system performance - state of the art

I get asked about this a bit. What latency, wire to wire, stimulus to response, would be state of the art for a trading system or EMS?

How quaint: 1 ms == 1,000,000 ns
Ask two people and you'll get seven opinions. This post represents my current meandering point of view. At least until later today, perhaps.

First, a little historical context.  A little after the turn of the century, I started a new job at Susquehanna. Someone had drawn with a marker the important goal for the team on the window. It simply said, "< 2ms." In those ye olde times of yore, the fastest vendor system going around was Orc's Liquidator which was not far from around 0.5 ms with its approximately seven figure price tag. Seems positively Penny Farthing now. Today we count nanoseconds.

Ignoring other factors, which are almost always more important, my view on state of the art engineering, assuming a 10G connection, is roughly:

Hardware
Ho-hum <  2,000 ns
Good <     500 ns
Excellent <     100 ns


Software
Ho-hum < 10,000 ns
Good <  5,000 ns
Excellent <  2,000 ns

Today, low latency is a necessary but insufficient condition for success. Sufficient speed is assumed. You have to dial in your peculiar "sufficiency" for your circumstance. Fast enough with an eye to cost as you can only win with trades you can afford to do. Low latency capabilities are now widespread and thus the main focus is typically elsewhere.

However, we all need to remember, Knuth's "premature optimization is the root of all evil" does not apply here. Architecture has to be planned. Too late is too late.

Waste not, want not.

--Matt.


The following Asset Managers need to select their tools with greater care:


Monday, 24 August 2015

ITG Posit - Aug 3 Finra ATS stats - was 62% higher before scandal

The first full week of trading after the announcement of a possible ITG SEC settlement has percolated out via the FINRA ATS Statistics for Tier 1 US stocks. The statistics are post the ITG press release but before the gory SEC details were available. The SEC's settlement announcement and details were published August 12.

The raw numbers for ITG POSIT are:

Date Num Shares ATS share%
3-August-2015 96,944,300 2.62%
27-July-2015 144,365,800 3.88%
20-July-2015 175,855,300 4.92%
13-July-2015 142,353,700 4.57%
6-July-2015 167,849,900 4.84%


The 96.9M shares reported for the week of August 3 has to be increased 62.5% to get to the average of 157.6M shares reported for the weeks from July 27 to July 6. This is a cute way of describing a 38.5% fall in the numbers of shares traded by POSIT compared to the prior weeks' average.

Below is a rather hastily put together chart of the history of recent ATS Tier 1 stocks (click on the picture to enlarge). ITG POSIT is the large orange line with the big circular symbols. IEX has picked up share as ITG has fallen and, to a lesser extent, so has MSPL MS POOL (ATS-4) as other ATS stats also lifted slightly.

ITG continues to trade significantly above its 52 week low and further board changes have not yet been announced.
ATS data is provided via www.FINRA.org/ATS and is copyrighted by FINRA 2015







Thursday, 13 August 2015

ITG - post settlement meanderings

Ex-CEO Robert Gasser's compensation from 2010 to 2014 was:

$15,695,178 = 77% of the SEC settlement

The SEC's findings suggest there may be a case for clawback.

Reflections in a Dark Pool (source)
Given the Securities Exchange Act of 1934 Section 17 violations already found and admitted, there seems to be a prima facie case for the same argument for the non-omega $70+M ITG has made from prop trading since 2002.

There is no direct evidence of interaction with customer flows from those trades, just hearsay and scepticism. Nevertheless, the same SEC argument applies: many buy-side customers were misled as to the agency-only nature of the firm directly by ITG, or indirectly via their sales and marketing collateral.

Perhaps further penalties would be lighter than the previous ratio of nine to one ($18M to $2M)? That would be for the best as otherwise any disgorgement of such profits would be a significant portion of ITG's June 30 cash balance of $211.79M.

I do wonder what interest really exists in the various ITG investor class suits. It seems likely to be more distraction than reality. The top 10 institutions own 46.36% but the median holding is just 0.015% of assets. This suggests the main shareholders have more important things to worry about. Perhaps aggrieved customers would be a better target for the law firms as a refund of trading fees may be of interest. Not quite the same cookie-cutter type of class lawsuit though.

ITG's insider sales list may be due for some scrutiny.

Kevin J.P. O'Hara's resignation from the board looks better today. 

The board allowed Project Omega to start. Why is the rest of the board not being held to account?

A sale or merger (reverse or otherwise) seems inevitable as a different level of trust is only possible with completely new management, and a new board.

It's not the end of the beginning but the beginning of the end.

--Matt.

ITG - Prop trading for June Q 2015, Y2014, Y2013 and down the rabbit hole

As I've nauseously over-meandered previously, here (Dec 2014), here (one line July 2015), here (Dead Firm Walking), here (Gasser Gone), and here (Settlement), ITG continues to proprietary trade.

Here are the rough figures from 2000 that I could find in the 10Qs and 10Ks. Generally it represents ITG Canada's "Other" revenue line (excluding some mentioned items):


$ in thousands
Q2 2015 1,770
Q1 2015 1,834
Y2014 8,526
Y2013 8,193
Y2012 6,020
Y2011 6,282
Y2010 5,572
Y2009 8,450
Y2008 - Peak? 16,512
Y2007 11,042
Y2006 8,800
Y2005 increased
Y2004 ?
Y2003 ?
Y2002 ?
Y2001 ?
Y2000 ?


Total 83,001

Roughly, ITG's non-Omega prop trading was worth somewhere between $70M and $100M. It is not clear though and $70M would seem too low. 2005 was the year I was made aware of it but it wasn't obvious from the financial reports back then. Perhaps there is more in other line items I've not touched on here.

So, quite a bit more profit than Project Omega's $2M from the SEC settlement. ITG was not so hopeless at prop trading that the limited $2M made them seem.

Here are a few quotes from the reports:
2008: "Interlisted arbitrage trading revenues improved to $15.9 million compared with $10.9 million in the prior year, benefiting from higher market volatility."
2007: "Other revenues included $10.9 million from our interlisted arbitrage activities versus $8.8 million in 2006"
2002: "(e) income/loss from positions taken by ITG Canada as customer facilitations (a customary practice in the Canadian marketplace) as well as income from same day Canadian interlisted arbitrage trading."
2009: "Revenues from principal trading (included in other revenues) were down in 2009 compared to 2008. Aside from an unfavorable foreign exchange impact of $0.6 million, the reduction was attributable to several factors including lower market volatility and the expanding presence of high frequency participants (which has narrowed spreads and therefore reduced profitable trading opportunities)."
2010: "Revenues from principal trading (included in other revenues) were lower in 2010 as the expanded presence of professional trading firms has significantly reduced the spread to be earned and thus limited principal."
In ITG's latest 10-Q filed with the SEC on August 10, 2015, on page 20 of the ITG report to the SEC,
"Other revenues include: (i) income from principal trading in Canada, including arbitrage trading,"
However, this quarter ITG is more circumspect than usual in the Canadian segment and fail to elaborate what the "Other" line item consists of in any regard. In previously quarters "Other" has typically been mainly prop trading via the US / CAD interlisted stock arbitrage, but there are no indications this time. If typical, prop trading would be most of the $1.77M "Other" revenue for the quarter.

In the 10-K for the year ended Dec 31 2014, we can see the further mention on page 31 and page 32,

with the following comment,
"Other revenues increased slightly due to an increase in income earned on foreign exchange transactions and principal arbitrage trading, partially offset by losses on client accommodations and the impact of currency translation."
So, it seems around $8M for both 2014 and 2013 for ITG's prop trading? Prop trading certainly may help the bottom line.

We don't know if customer or pool activity is involved directly or indirectly for stock or FX as part of this prop arb.

At least in the 10-K on page 12 with regard to risks we can read,

"We incur limited principal trading risk in our Canadian Operations. 

        A limited portion of our revenues is derived from principal trading in our Canadian Operations, including arbitrage trading and the net spread on foreign exchange contracts executed to facilitate equity trades by clients in different currencies. As a result of this trading, we may incur losses relating to the purchase or sale of securities and currencies for our own account. Although we attempt to close out all of our positions by the end of the day, we bear the risk of market fluctuations and we may incur losses due to changes in the prices of such securities and currencies. Any principal gains or losses resulting from these positions could have a disproportionate effect, positive or negative, on our revenues and profits, and could also result in reputational damage."
Maybe ITG should stop prop trading?

--Matt.

_____________________________
PS: I wonder if any of these insider sales may be problematic?