High Frequency Trading vs Flash Trading

Economist Tim Worstall has an distinction to make on the differences between high frequency trading and flash trading in a recent article.

Essentially it is the difference between getting your orders in quicker than every one else, and having a peek at what everyone else is doing before putting your money down.  The SEC appears to be conflating the two and has concerns.

With the world condition in banking, could we see some poorly thought out legislation rushed through so that regulators can be seen "doing something"?  Or would it level the playing field a little so that those trading operations that cannot afford the overhead of super fast computers and networks are not excluded?

Tick Size Harmony...

...in a rare show of co-operation (I wonder what is the carrot or (regulatory) stick here to motivate this?) European exchanges and MTFs seem to have agreed on standardising tick sizes (or at least to have two standards rather than twenty five!). Extract from article on AutomatedTrader:

"From the perspective of each trading venue, strong incentives exist to undercut others in terms of tick sizes, which is not in the interest of market efficiency or the users and end investors. This might, in turn, lead to excessively reduced tick sizes in the market. Excessively granular tick sizes in securities can have a detrimental effect to market depth (i.e. to liquidity). An excessive granularity of tick sizes could lead to significantly increased costs for the many users of each exchange throughout the value chain; and have spillover costs for the derivatives exchanges' clients."

Best execution 2009 - July 1st 2009

A few summary points I took from the Best Execution Europe 2009 event courtesy of Incisive Media that I attended yesterday morning.

The event started with a presentation by Michael Fridrich, Legal and Policy Affairs Officer of the European Commission:

• From what Michael was saying then in my view, it seems that the EU is using the G20 declaration on financial stability in April as a remit to regulate in many areas (not all of which related to the current crisis, see last paragraph in this post)
• He said that the EU is currently working on removing national options/discretions with respect to financial markets in order to create a single EU rule book and combining this with stronger powers for supervisors including much harsher sanctions against offending institutions
• They are also reviewing the necessary information provided to investors in OTCs, even if the investors qualify as "professional investors" under Mifid.
• The EU is currently reviewing Mifid and the Market Abuse Directive (called "MAD" which is at least humorous...)
• EU is also unsurprisingly looking at the regulation of Credit Ratings Agencies (CRAs) given their involvement in rating CDOs and other structured products

So in summary it was a civil servant PR exercise with few surprises, other than we are going to regulate anything that moves. On to a panel debate on "build vs. buy" for execution management software. I will try and put my obvious vendor bias to one side in summarising this one:

• The panel summarised that this decision was about the usual issues of time to market and what is an institutions core IP
• A senior IT manager from JPMorgan said they both build and buy - but given the size of their organisation and the need to innovate they do build a lot
• The COO of Majedie Asset Management said that "build" was "20th Century" and the IT should focus now on "assembly"
• He added that if IT lead a procurement process he finds this tends to lead to more proprietary solutions than if business is managing it.
• He summarised that business people should have the mandate to define inputs/outputs to a requirement and that IT were not qualified to do this.
• Putting it more controvertially he suggested that IT people should work for IT companies
• The JPMorgan guy responded that "assembly" of external components can lead to excessive staffing in managing all the plumbing, and that build in house could build a more generic and targetted platform that would need less management
• The moderator summarised the build vs. buy decision as one of balancing time to market and how bespoke a solution is alongside of looking at the risks for buying of 1) integration risk 2) vendor risk and for building of 1) delivery risk 2) key man risk

The debate on this was pretty standard, but the guy from Majedie was at least controvertial in what he was saying, (including at one point that "investment management does not scale"). I assume he is trading simple products and as such is able to outsource more than the JPMorgan manager. My own slant is that more vendor products need to be designed to integrate easily with the IPR of a financial institution i.e. less black box.

Tom Middleton of Citi then did a presentation on (equity) market liquidity and market fragmentation:

• He started by saying the Smart Order Routing (SOR) was like "Putting Humpty-Dumpty back together again" from all the sources of liquidity now available under Mifid.
• Being no expert in SOR, I was excited (?) to learn a new term which was "finding Icebergs" - apparently an "Iceberg" is a large non-public ("dark")  order being posted with a much smaller public trade order.
• He said that market fragmentation will increase further but there will be less trading venues as the market consolidates.
• New algorithms will be developed more specifically for trading on dark pools of liquidity
• Clearing and settlement costs are still high across Europe which limits the usage of small size orders in trading but trading volumes will continue to grow
• The drive to ever-lower latency will also continue
• Usage of SOR will grow

Tom's presentation was then followed by a panel debate on Smart Order Routing:

• A manager from Baader said that the German area market of Europe was not very sophisticated yet, with most German clients specifying exactly where the trade should be executed hence nullifying the need for SOR.
• Deutsche Bank (DB) mentioned that having both US and EU operations had helped them get SOR in place for the EU quicker given their US experience.
• UBS and Baader both said that Algo trading and SOR are increasingly integrated and will merge with the Algo define what and how to trade and the SOR component determining where
• DB said that a "tipping point" towards usage of SOR in the EU will occur when more than 20% of trading occurs away from the primary exchanges.
• DB said that 60% of US liquidity was due to algorithmic trading and that there were now no EU barriers to this happening in European markets and bringing with it increased liquidity, although issues such as not having a consolidated market tape for trading made things more difficult
• Neonet said that clearing and settlement costs were still a barrier to widescale SOR adoption.
• IGNIS Asset Management said that SOR was a "high touch" service for them, requiring SOR vendors to be very responsive and client focussed. In selecting SOR vendors they were concerned with data privacy and also with having a real-time reporting facility to see how orders were being filled.

And finally (at least before I had to leave) there was a presentation by Richard Semark of UBS on Transaction Cost Analysis (TCA):

• He was surprised to find that there were not many presentations around on TCA
• TCA vendors are behind the times and are not up to date with current developments
• Historically TCA was about what had happened (about 3-4 months ago!)
• Mifid has driven fund managers and traders to talk more and TCA is a key part of this conversation
• It is hard to look bad against traditional TCA measures such as VWAP if a stock is always rising or always falling, and this can hide a lack of performance and "value add"
• Using "Dark" for non-displayed liquidity has been a publicity disaster for the electronic trading industry
• Much Smart Order Routing (SOR) is still based on static tables of trading venues that are updated on a monthly or quarterly basis
• Market share by volume of a venue is not necessarily correlated with obtaining the best prices in the market
• TCA should be based upon a dynamic benchmark that responds to the market and trades done not against a static one
• Trade performance is not linear with trade size which is an incorrect assumption in much of TCA
• Trade risk (variability in outcomes) deserves more focus
• Portfolio TCA is much more complicated where the trading of a single stock cannot be looked at in isolation of its effects on the whole portfolio
• Real-Time TCA is becoming ever more important to clients since it allows them to understand more of what is going wrong/right with filling an order
• TCA providers are not doing a good job for clients, not using the right data or answering the right questions for clients

Not sure who the TCA providers he refers to are, but maybe I should find out to see what they offer...

 

 

 

 

Twittering the Wisdom of Crowds

Deserving an award for title alliteration, an article on Finextra has announced that Streambase Systems have connected their system to Twitter, the fashionable microblogging site. Regardless of the intent, it is an excellent marketing exercise by Streambase (er, maybe one that I should remember for the future!...).

Reasonable comments from Finextra at the end of the article, saying that Twitter is a notoriously bad source of information, very open to (designed for?) rumour, and as such it would be difficult to see what real information traders could extract from the noise. At one level, then rumour and counter-rumour are the basis of markets, although the recent financial crisis has illustrated how powerful rumours can be. I would suggest it begs the question as to when rumour and counter-rumour is part of the price formation process, and when it becomes market manipulation.

On a related note, the Efficient Market Hypothesis (EMH), the financial theory that all information (including rumours) is reflected in current prices, has been coming under some attack in the press recently. With a fund-management and Monty-Pythonesque slant, James Montier of Société Générale takes EMH to task in his recent article in the FT (see Pablo Triana for an alternative view).

My opinion is that EMH has still got some legs in it as a model, but behavioural finance probably has a lot more to explain (or rationalise?) about this theory and others in light of recent events. Anyone got a different opinion, or do I need to open a Twitter account to find out?...

Alternatives Need a Bigger Umbrella?

Interesting article in the FT today about why the US exodus from traditional exchanges might not be repeated here in Europe, which is contrary to the recent marketing mantra of the alternative trading venues such as Chi-X, Turquoise and Equiduct. If correct, the economics outlined in the article look justifiably prohibitive:

"Merely to break even, an alternative platform with a cost base of about €10m would need to do 100m trades a year. Quite a task, given that the 208-year-old London Stock Exchange, which reports full-year figures on Wednesday, said in March it was on course for about 190m in its UK orderbook."

The article points out the difficulty of starting an alternative trading venue against a dire economic background and emphasises this by ending with:

“Xavier Rolet, the LSE’s new chief executive, should be praying for rain.”

Microsoft CEP Surfaces as "Orinoco"

Seems like Microsoft have now gone public on the Microsoft TechEd site that they have a Complex Event Processing (CEP) engine that will be coming to market shortly (see MagmaSystems blog post ). One of my colleagues Mark Woodgate attended a briefing event at Microsoft for this technology back in February this year - here's an extract from some internal notes that Mark made back then:

"Microsoft CEP is very similar to StreamBase conceptually (and not unsurprisingly), in the sense that there are adapters and streams and how you merge and split them via some kind of query language is the same. However, StreamBase uses the StreamSQL which as we have seen is SQL-like in syntax but Microsoft CEP uses LINQ and .NET and although conceptually it is doing the same thing, it does not look the same. StreamBase’s argument was you can be an SQL programmer to use it and don’t need lower-level like .NET; however, it’s not SQL really as it has all these ‘extensions’ you have to learn so using .NET might look more tricky but in fact it makes sense. They don’t have a sexy GUI yet for designing CEP applications like StreamBase but it will be done in Visual Studio 2008.

 

Currently, you build various assemblies (I/O adapters, queries and functions) and then bolt them all together, called ‘binding’ by command line tool. You then deploy the application onto one or more machines using another tool so it’s a manual process right now. They are aware this needs to be made easier and more visual. They are allowing other libraries to be bolted in via the various SDKs so it’s pretty open and flexible. It works well with HPC and clusters/grids (or so they say) and of course can be used with SQL Server. The CEP engine also has a web interface based on SOAP so at least non-Windows based systems can talk to it"

 

The release of this technology will be an interesting addition to the CEP market and to the Microsoft technology stack in general. Assuming performance is at credible levels (i.e. not necessarily leading but not appalling either) it will certainly bring both technical and commercial pressure to bare on existing CEP vendors (see earlier post on Aleri/Coral8) and has the potential to broaden the usage of CEP. Obviously Linux-Lovers (sorry, I didn't mean to be personal...) will not agree with this, but Microsoft is putting together an interesting stack of technology when you see this CEP engine, Microsoft HPC and Microsoft Velocity coming together under .NET.

 

Merging in public is difficult...

Sounds like Aleri and Coral8 in the CEP (Complex Event Processing) market are not doing the best job they could of managing the publicity surrounding their recent merger, not helped by announcement of a CEP capability by Sybase, based on Coral8 source code. 

Explained more in a post on the Magmasystems Blog, and made more entertaining by the aggressive marketing tactics of Streambase in responding to the merger by offering a software trade-in facility for clients of Aleri and Coral8 (see press release).

CEP in 2009

Interesting predictions for complex event processing (CEP) in 2009 (click here for link) - sounds like some form of reality is appearing in this area of the market, accelerated by the current financial crisis. Entry of bigger players and usage of LINQ in CEP will be interesting too.

Never ending liquidity for FX?

FX volumes grow from $99,000bn in 2007 from $71,000bn in 2006 - growth driven by automated trading, cheaper execution leading to more and more participants. How fluid can the market become? Article link:

http://www.ft.com/cms/s/0/10fc4512-2c4f-11dd-9861-000077b07658.html?nclick_check=1

Vhayu and Streambase - positioning clarified?

Partner announcement on Finextra with Vhayu and Streambase coming together:

http://www.finextra.com/fullpr.asp?id=21477

Defining what vendors mean by a "Data Management System" is difficult enough for clients, but in the area of the somewhat fuzzy technology definitions around automated trading it is interesting to see Streambase clarify their offering around CEP (and not database too, which was one of their first messages around bringing real-time and historic data together), and that Vhayu seems to be emphasising its tick database capabilities (and de-emphasising its original perception in the market as a CEP vendor).

TradeTech Paris 2008 - Chi-X leads the waiting game

We exhibited at the TradeTech Paris event last week - this is a mainly equities/trade execution event and as such most of the speakers were playing the waiting game and not saying much. Everyone seems to want to see more of the new trading venues come on line before they put out any opinions of substance.

The main (only?) news item was the rapid uptake of Chi-X and its share in trading volumes against the exchanges. There was some "competitive" banter between Peter Randall of Chi-X and Eli Lederman of Project Turquoise (Chi-X saying that Chi-X is a live business and not just a "project"). LSE was there too, not doing a great job of defending the record of exchanges - too much emphasis on defending their existance based on the past rather than the future in my opinion.

Streaming Blue Genes...

The supercomputer continues to make a come-back - just up on Finextra with TD Bank testing IBM's Blue Gene supercomputers to amalgamate and analyse real-time structured and unstructured data:

http://www.finextra.com/fullstory.asp?id=18293

Light speed relief from Latency marketing...

Just been reading a report on latency from Inside Market Data - main thing that hit me about the vendor discussions in the report is that they seem to be finding it more and more difficult to differentiate their offerings (aka: come up with much new/interesting to say) when talking about latency and approaching "light speed" physical limits on speed of communication. The Latency discussion seems to be getting commoditised and overdone from a marketing perspective, even if the problem is a real one with money to be made (and spent!).

Best comment comes from the editor Max Bowie, who suggests that (maybe ironically given the content of the report?) that speed is not the only factor behind successful automated trading, echoing something I heard recently that "...if your core trading strategy is only based on being faster at trade execution than the likes of Goldman Sachs, then get a different job...". The "Latency" special report can be found at:

http://www.insidemarketdata.com/public/showPage.html?page=imd_reports

Far away from low latency...

...is the OTC derivatives market - still a lot to be done according to Finextra report and letter from the Fed:

http://www.finextra.com/fullstory.asp?id=18285

Co-location for Co-location?

Co-location has obviously been very popular in bringing down the latency of trade execution, with algorithmic trading software being located on-site at trading venues by the banks and hedge funds. Interesting problem brought to my attention by a senior equities technologist at one of the banks is that co-location is fantastic at reducing latency if you are considering just one trading venue.

In the new(ish) world of multiple trading venues then regardless of how low the latency is from co-location at one venue, centralised smart order routing (SOR) means that these advantages are lost to a degree since you always have the latency of the round-trip on market data and execution to consider to and from the centralised SOR facility to each of the separate (remote!) trading venue. Unsurprisingly then many of the players are looking at locating SOR and algo software physically near multiple trading venues, Makes you wonder where it will stop, maybe all of the exchanges in the world should relocate to somewhere like Slough?...

Chi-X on "Old-School Exchanges"...

Peter Randall of Chi-X did a good (well, both aggressive and funny in parts...) presentation at the Intel Faster Trading Event - he put forward the LSE as an "Old-School" institution, saying that in taking trading electronic (years back) the LSE simply computerised the existing human interaction, using software to implement a good(ish) matching engine.

He then said that today's focus should be on the I/O queue (trade transaction throughput) since he suggested that many brokers have been asked by the exchange "would you mind slowing down your trading, old boy" as trading algorithm throughput becomes limited by the capacity of the exchange. He mentioned the LSE has a capability of 9,000 transactions per second whereas Chi-X has around 30,000 tps.

Peter said that the Exchanges (and Regulators) didn't like the "f-word" (steady!...) of "fragmentation" (i.e. moving away from a national exchange (concentration) model). He them got a little more fruity talking of how the exchanges also hate the "c-word" (where is this going?...) with "competition". He then justifed Chi-X's current standing by listing off for example that Chi-X did 9.69% of the trading volume in the FTSE yesterday and said that Chi-X typically offering 2.05 basis points cost improvement over trading on LSE. Key points in their success where the ready use of FIX as an industry standard, plus their approach of giving market data away for free (already appearing on Bloomberg terminals for instance).

Taking some lessons from the US, he said that two key points in ensuring the future success of Chi-X (and other MTFs) were to "equitise flow" (i.e. get the major institutions who use Chi-X to also become equity investors) plus he said that a venue needed to offer very low cost to institutions that make and take trades in the same security during the day. Good presentation overall, certainly very clear who Chi-X is attacking...

Credit Suisse - Algo trading pitch

Went along to the Intel Faster Trading event last night - overall a good event, interspersed with the usual buzzword bingo from some of the vendors presenting/talking! First up was someone from Credit Suisse, fundamentally doing a sales pitch on their algorithmic trading capability to the buy side. Some points on MiFID:

• MiFID fundamentally aimed at competition and reducing cost of trading
• Median trading spreads in Europe about 5x US median
• Median trading latencies in Europe about 18x US median

Put forward that there would be around 10 alternative trading venues in place by 2009 (seems low?). Existing MTFs included LiquidNet (received +v feedback on them from NY contact), ITG and Chi-X. Key point was that Credit Suisse said that LiquidNet and ITG had so far excluded sell-side brokers and hedge funds, so as to increase confidence in the venue amongst the buy-side (in the absence of "predators"!).

Credit Suisse then listed off planned MTFs including BATS Trading, Plus Markets, Equiduct, Turquoise, Euro Millenium, Open Match and interestingly (given it is an exchange not a start-up MTF!) NYSE Euronext with their DarkPool offering.

Big pitch on Smart Order Routing (SOR) capability, saying that Credit Suisse averages a saving of 1.7 basis points (to I am not sure what, probably exchange/single venue trading). Interesting that they say the important focus is moving from low latency to transaction processing speeds (echoing some earlier feedback on the focus moving to getting shorter message lengths to increase throughput).