There has been some recent discussion / warnings (again) about the perils of the Carry Trade and how it is distorting the global markets by cheapening the Yen.

The Economist just ran an article on Carry Trading, ”Carry On Speculating“, and previously “Carry On Living Dangerously“ (their titles playing on the name of the trade and the classic Carry On British comedies of the 60s and 70s).

Basically, the carry trade is where an investor borrows money cheaply and invests it in instruments that have higher returns. Often the borrowed money flows into higher risk assets (such as emerging markets), so there are clear dangers in this activity. However, the currency carry trade is even more dangerous since it involves borrowing in one currency (that has a lower set of interest rates) and using that to invest in assets in another currency, so if the currency rate moves against you, your assets will be devalued.

The weird thing though about the Carry Trade is that is goes against economic theory. Normally the FX market displays “covered interest parity” whereby the difference in the currency spot rate and forward rate reflects the difference in the respective interest rates. This in turn would represent an investors expectations that the low interest rate currency would appreciate against the higher interest rate currency. However, since investors are borrowing the lower interest rate currency and buying assets with a higher return in another currency, and hence SELLING the borrowed low interest currency, the carry trade actually applies DOWNWARD pressure on the borrowed currency. As long as low interest rates exist, the currency everyone is borrowing gets key pinned down, and hence more carry trades get done, keeping the cycle going.

Almost all of the recent discussion has been specifically around carry trading in Yen, which has very low interest rates and is pretty much the basis for the global currency carry trading, and hence global hyperinflation. But it is tough to know how much investment there is in carry trades. For example, hedge funds do carry trades through ccy forward swaps which are off-balance sheet. Estimates,  based on the record net “short” positions in yen futures on the Chicago Mercantile Exchange, suggest the total size of the carry trade range as high as $1 trillion.

So, Yen is being cheapened and the risk in global markets continues to increase. If the Bank Of Japan raises interest rates (which they have made noises of doing), the whole carry market could be unwound through the sell-off of the assets that were funded by borrowing all this cheap Yen. According to some reports, it would take a 2% increase in interest rates to undermine these carry trades, but many people believe this is not going to happen any time soon, so they keep doing the trades and the cycle continues.

At some point though, something will happen. And the more the Yen slides (because of the selling of Yen), the more likely it is there will be a sharp recoil.

The next few months will be interesting to watch. Investors beware, you mind find yourself in one of those other Carry On comedies, such as “Carry On Up The Kyber”. Or maybe “Carry On Matron” ?!

This presentation was given by Mike DeSanti of Eikos Partners (a partner of Lab49) — he was supposed to do this with a Microsoft guy, but as Mike said “the guy from Microsoft had a virus and could not make it”. Very drole.

Mike ran through some standard Microsoft slides on the .NET 3.0 framework covering:

• Windows Communication Foundation (WCF)  - increases productivity and promotes service-orientated development through  support of standards, message-orientated programming, attribute-based development, configuration-based communication, etc.
• Windows Presentation Framework (WPF) - lots of cool UI tools and components, supports declarative programming that includes XAML (Extensible Application Markup Language)
• Windows Workflow Framework (WWF)  - supports transparent declarative model-driven software,  implementation of long-running or stateful processes as well as flexible control flow (sequential, state machine, rules-driven)
• Cardspace

Mike did a good job of pointing out where the various elements of the software stack can be applied to SOA. Our friend DLG was there to help out as well on some of the more detailed questions from the audience. But I thought Mike’s own slides on the trends in Financial Services were more interesting and indicative of what is actually going on in financial services. Key points:

Sell Side Trends & Issues

• SOAs tend to be islands of implementations and specific web services tend to be organized along specific business or organizational units.
• There is quite a bit of refactoring going on of (legacy) applications, and in some cases messaging buses are being used for pre-web standard applications to communicate into the architecture.
• There is a general lack of co-ordination around SOA development
• Business Process Layers are NOT being used to orchestrate Services. This is an issue given the complexity and the statefulness of some of these processes.

Buy-Side SOA Trends & Issues

• Little investment in SOA and where they are trying to implement SOA, there tends to be a loose conformity to SOA standards/guidelines.
• Business services tend to service multiple asset classes
• Buy-side firms tend not to model their business process very well and tend to standardize on a single SOA solution stack (such as Microsoft!)
• The services infrastructure is generally lighter as it does not need to handle as large volumes and in many cases does not need to support the complexities the sell-side firms experience.
• There is has been a lack of broker-dealer standardization, so it is tough to reconcile services and data.
• But they are beginning to use SOA technology and workflow models to integrate legacy applications and to buildreconciliation systems to manage multiple broker-dealers.

Overall, it would seem that Microsoft technologies are already being used to address many of the issues in Financial Services and that .NET 3.0 will prove to be a great enabler going forward. Compared to the IBM stack, the new Microsoft software felt much more in-tune with defining and developing compelling service-driven applications that support business workflow processes and very rich performant front-ends.

Well, once again, IBM leads off another conference with its keynote speech on how they have already solved the entire world’s problems. Although you know how the presentation is going to pan out, I was pleasantly impressed with the speaker, Sandra Carter, VP SOA and Websphere Strategy, Channels and Marketing — she stood in front of the podium without any prompts and delivered the presentation in a very professional and confident manner.

The IBM pitch was not a bad one and pointed out some of the considerations for building an SOA, such as being able to simulate and adapt to changes in the business process model. That said, there were some pretty strange statistics being quoted such as: “74% of CIOs with SOA strategies are on the executive committee of their company, versus 59% of CIOs without SOAs are not”, and “the average salary of a CIO with an SOA is $250k, versus $159k for those who do not”. That second statistic did not sound like Wall Street to me.

When IBM got into the details, there were all kinds of tools (and acronyms) being thrown around: Component Business Model (CBM), Service Orientated Modeling and Architecture (SOMA), SOA Business Catalog, SOA Maturity Model, Assessment Tool and Workshops, etc. On the product side of things, they illustrated a couple of case studies where there were different drivers for adopting SOA, each of which utilized a different set of various Websphere products such as Websphere Business Services Fabric, Process Server, Message Broker, MQ, DB2, etc.

Some time was then spent on Websphere Front Office for Financial Markets which is a “sharable low latency market data distribution service” - that sounds more like a repackaging of MQ to me than any revolutionary new product offering.  It also has “Reliable Multicast Messaging”, which is an upgraded pub/sub mechanism with higher performance levels. Additionally, “it prevents loss of market data ticks” and has metering, monitoring, auditing and entitlement capacilities. Specifically, this *new* piece of infrastructure can handle 100,000 messages/second and a latency in the order of 300 microseconds. It can also hook up to various market data providers such as Opera and Bloomberg (who are a development partner for this product) using their “binary self-describing ticks”. Sounds to me that IBM have decided that market data is their new strategy for financial markets. This was all demonstrated in an application that had a LOT of data flashing rapidly across it (we could not figure out if the demo was written in flash or Winforms - either way it did not look like a trading application !).

The presentation ended with a quick run through of thier Enterprise Service Bus Offering that includes 3 products for SOA: Websphere ESB (for web service integration); Websphere Message Broker (for web service and non-web service assets); and Websphere DataPower (a specialized hardware/software appliance that provides performance acceleration on XML processing as well as enhanced security capacilities).

In summary, it was typical stuff — lots of plumbing that probably needs a lot of IBM consulting services to generate a lot of reports/assessments to tell you how to use it.

Since a couple of us at Lab49 have been working on architecture projects for clients, I thought it might be useful to attend the WebServices / SOA On Wall Street conference to see what the usual plethora of vendors had to say for themselves these days. Not much is my verdict.

In the mostpart, this was attended by vendors who are selling a lot of infrastructure that has been given an SOA-spin. From integration servers to messaging infrastructures to testing frameworks to SOA Governance solutions to systems monitoring and management solutions. Quite a few of the vendors were clearly not established in the FS space (e.g. a company called RTI is predominantly an Aerospace and Military provider of a low latency high throughput messaging solution  — that happens to run all the messaging on US Navy warships– but spun as a market data distribution platform).

As usual, IBM and Microsoft were the elephants in the room, touting their soup to nuts software stacks for SOAs  (more on these in other postings).

I was disappointed that no-one had any real practical insight of how to actually build business-critical applications that run on an SOA. Yes, there were broad examples of the results of some client work, but nothing that touched on the nuances and complexities of actually building these solutions — other than IBM’s “web services is the industry standard best practice for implementing SOAs” ! No-one was talking about the types of the issues we have to deal with on a daily basis such as loose vs tight coupling, event-driven messaging, synchronization, business process control, migration of legacy applications, etc

I see GemStone has announced that it now has support for native C++ and .NET clients to access their distributed caching product, Gemfire. Will be interesting to see what the uptake of that is on the desktop of traders, relative to their competitors in the marketplace. Unlike the products of their competitors (Tangosol  and  Gigaspaces) which are written in Java, Gemfire’s enterprise product actually comes in two flavours, Java and C++.

Meanwhile Tangosol, have recently announced support for .NET applications to use their Coherence data cache, they now also support Spring to allow the management of the data cache.

Not to be outdone, Gigaspaces also recently announced support for .NET and “introduced the concept” of PONOs (Plain Old .NET Objects) to allow consistency of programming models across Java and .NET worlds.