The Windows HPC Team Blog : Actuarial Sciences

Matlab Users among Actuaries?

daved18 — Fri, 07 Nov 2008 03:29:00 GMT

We recently had the chance to speak with some modelers using Matlab to run some analysis for their Enterprise Risk Management program.

They were surprised to find out that Matlab was supported on the Windows Compute Cluster Server. Is anybody else out there running Matlab for Economic capital analysis or ERM?

Please let me know,

Why choose a single tool?

daved18 — Sat, 27 Sep 2008 13:22:00 GMT

Last week at the Valuation Actuaries Symposium I was able to attend a session hosted by Milliman at the Embassy Suites Hotel during which the MG-ALFA modeling product and the MG-Triton Valuation product were discussed and reviewed in the context of how they contributed to their combined effectiveness.

One of the questions raised during the session focused on the possibility of using a single modeling tool for both deterministic and Stochastic modeling. My interpretation of this question was , would it be possible in the future to have a single modeling tool for all purposes?

My reaction to this question was, why would we want to limit ourselves to a single tool? It might be easier to provision and train your team on one tool, but in an increasingly specialized world it seems that we would be missing out on a lot of expertise.

As an audience member I had the opportunity to add comments to the answer on the question. All I could say is that as a representative of Microsoft, I felt my way to add value to this community is by working to make all the tools for actuaries available on a single easily accessed and flexibly deployed platform.

I'd love to hear your opinions on this subject. Is it a worthy goal to get all the tools on a single compute platform?

Are Actuarial clusters becoming a real possibility?

daved18 — Mon, 08 Sep 2008 08:39:00 GMT

In the past couple of years I've learned just enough about how computational modeling is used in Actuarial science to begin to understand the real possibilities.

In a nutshell, I've found the big three applications are general capital model for purposes ranging from pricing to cash flow testing. The second primary application is policy valuation, which appears to be gaining momentum as more than a regulatory requirement, becoming a basis for capital modeling based on actual experience and valuation at the current time. The third computational demand is from the operation of hedging programs to mitigate the risk to a company from the guarantees associated with Variable Annuities or Equity linked annuities.

As policy valuation systems grow to include functionality that dramatically increases compute requirements it will join Hedging and Capital Modeling as a consumer of massive compute cycles increasing the demand for shared computing resources.

I May be a lay person when it comes to Actuarial modeling, but I do have over 20 years experience when it comes to designing computer systems and Infrastructure. I know that combining these three types of activities onto a single computing grid has a number of benefits, including:

More efficient use of computer hardware
Synergy with a the predictable business cycle of Actuarial modeling, to allow shared use that balances out demand on a cyclical resource.
Flexibility in responding to expedited demands for pricing or profitability analysis

In the past it's been impossible to combine these applications into a single Gird or cluster because each application type used a different method for distributing work across the many systems required to return results in a reasonable amount of time.

As I write this all the major Modeling packages used in North America, MG-ALFA, TAS, Prophet, Axis, and Moses are for the first time available on a single scheduling platform. Windows High Performance Compute Server.

Stay tuned for more information about Policy Valuation applications and Hedging applications joining them on HPCS.

Once all the leading suppliers of the major application areas are supporting a single platform, the only thing between the actuary and enhanced computing capabilitiess is the completion of the IT Project to build a single unified Actuarial modeling cluster.

I can't wait to help make this happen.

Stochastic for Health and Property/Casualty

daved18 — Thu, 12 Jun 2008 10:23:00 GMT

Stochastic methods have been well integrated into the Life insurance segment of the industry. From my naive perspective HEalth and to a lesser degree P&C have been unable to benefit from these methods due to a lack of commercial software that exploits stochastic probability model.

I received an email from MArk Zanecki of IHA consutants describing a new software package that supports the introduction of Stochastic methods for Health and P&C. Mark takes advantage of Windows Compute Cluster Server to easily extend the required computation to multiple computers, significantly shortening the time to solution and meaningful answers.

Marks email is Quoted below

The aim of all insurers is to enhance the flow of information within their walls to be faster, more accurate, and through software/hardware at overalllower cost. Looking into the future, an insurer (health or property&casualty) requires actionalble information to be incorporated into pricing (notably trend), underwriting and forecasting for internal and external audiences. Looking at the past, the insurer requires accurate financial reporting, mostly notably with respect to reserve estimation for a variety of audiences including regualtors, tax authorities and financial rating such as AM Best, Moody's, S&P and Fitch etc.

IHA Stochastic Toolset for actuaries, built on HPC, fulfills this mission and frees the actuary to analyze results rather than developing the results. The aim ot the tool set is better quantification of risk to include measures or risk variability as required by actuarial profession bodies in US, EU, Canada and Australia. Moreover the actuarial value addivity to an insurer's internal flow is significantly enhanced as view from Chief Financial Officer as results are available in compressed time built on a auditable/reproducable process.

Too Much Marketing Hype?

daved18 — Fri, 06 Jun 2008 11:50:00 GMT

On May 28th I made my first SOA presentation at the Spring Health Conference. I was one of three vendor representatives on a panel covering the Actuaries High Performance Computing Challenge.

I thought I gave a reasonable presentation, but after the fact I received some feedback that it was bit Vendor centric and contained too much product information given the objectives of the conference. This was great feedback and I hope to use it to improve the presentation scheduled for the Life conference in Quebec on June 17.

If you plan on attending the Life conference please let me know what you want to hear about. My presentation at the health conference was focused on cost savings, because frankly, with the exception of the work by IHA consultants I had not been exposed to a great deal of computational modeling work by Health focused actuaries.

The Life Conference should be much different. The challenges of PBA, VA hedging and increased emphasis on adequacy of Economic capital have driven compute requirements up by up to 3 orders of magnitude. The presentation in Quebec should focus on the problem of using existing compute resources more efficiently to meet these requirements.

Please say hi if you see me in Quebec. If you mention the code word "Copper" I will have a small gift for you :-)

See you in Quebec

See You in LA on May 28-30

daved18 — Thu, 22 May 2008 11:52:00 GMT

The Society Of Actuaries annual health Spring meeting is scheduled for the Hyatt Regency in Century City next week.

I'll be presenting in Section 17 1:30 - 3:00 PM along with John Powers and Gordon Jackson.

Stop by and introduce yourself. If you mention the magic codeword "Copper", I'll have a special gift for you.

See you there

Help Me Help You - What Modeling packages do you need?

daved18 — Mon, 12 May 2008 12:25:00 GMT

The Windows Compute Cluster Server (WCCS) has come a long way in building value for the Actuarial Community. After about a year of active communication a nice little ecosystem of modeling tools has begun to build around WCCS and Windows Ccmpute Cluster Edition (WCCE).

As we look forward the most important question in my mind is "Where do I invest my time and energy to add value to this ecosystem". In other words what modeling tools do you need to get your work done. What modeling tools would benefit from access to a shared cluster? What modeling tools could use more support form Microsoft for scaling out computational capabilities?

As an example, MG-ALFA, Moses, TAS, Prophet, and AXIS all have announced or may be looking at ways to use the capabilities of WCCS or WCCE to improve their environments. What would be the best additions to these packages.

Is policy Valuation an area that needs more computational modeling support. How about catastrophic event modeling?

I know of at least one emerging consultancy interested in bringing stochastic methods to P&C and health in order to drive a reduction in the quarterly close cycles.

I could really use your comments. Please feel free to post your public comments to this blog or send me your private comments at my email address

Thanks

Dave Dorfman ddorfman@microsoft.com

Safety in Numbers

daved18 — Fri, 11 Apr 2008 17:16:00 GMT

Safety in Numbers?

Welcome to my first ever Blog entry, I’ve resisted entering the Blog-o-sphere up to now, but who will speak up for the actuarial users of Windows HPC tools if not me?

My first Post title is pretty clearly a play on words in a number of ways:

1) Actuaries are the people we rely on to keep us safe and meet our commitments by using numbers to understand the full risk profile. Somebody has to figure out what we can count on in the worst case scenario and our friends in the Actuarial sciences keep us safe with numbers.

2) The industry places a very high premium on reputation. So…. By nature the Actuary needs to be somewhat cautious in the methods they adopt. For this reason it’s important that the community know that many members will be relying on Windows HPC tools. A larger number of users equals a safer group of users.

3) With Wall Street leaving something to be desired in the area of full disclosure and the mortgage industry creating a bit of concern about the proper valuation of the underlying assets for CDO’s, there will probably be a strong demand to return to safe and reliable numbers to predict the stability and risk of an investment. Justifying how a number was arrived at will become increasingly important, and the documentation of methods to determine risk will become much more interesting. A well documented number is a safe number.

4) As the reverberations of the current financial scenario play out, focus will shift from enhanced returns on investment to reliable returns on investments. The credibility of the Actuarial community could become the new foundation of what has been called the Shadow Banking system. People who place faith in reproducible numbers are the safest people to put in charge of our investments.

I spend most of my day looking for Actuarial problems, finding the owners of those problems and explaining to them how Microsoft is interested in making it less expensive to deploy more computers in the execution of stochastic models to better understand risk.

So… If you know any actuaries, point them to me. One thing I’ve learned is that more computers is a good thing, allowing more exploration of the possible outcomes and more

Safety in numbers.