Hi,What are the systems people prefer for the risk management of derivatives in a hedge fund enviroment. Pretty much in the dark on this one, so any ideas suggestions, opinions much appreciated. Something that will take in account most instruments though. thanks

just, above all else, avoid "Murex", use the search function for my prior comments

Algorythmics is pretty much the market leader. It's not too bad either. The core riskwatch is pretty good and only riskmapper really sucks (so you write your won PERL routines for this part). overall definitely worth a look.If you're not so large had you though of having a a bureau do this for you? Have seen a couple iof settlement specialists who offer EOD risk as an optional extra.

Depending on the size of your company it's always better to create your own in house.

Why?

Marine is right, to the extent that buy vs build is a function of your size and requirement complexity, so there is usually not the same optimum for everyone.If you need a lot of functionality, then you will end up writing a lot of the system anyway. Thus the first question may be how easy the system is to customise, coupled with whether you are tied to the vendor for these changes.At some level, you have to stop, and buy black boxes. This may be a trade reconcilliation module, regulatory reporting, bond pricing, or low level memory allocation. The trick to to know where to stop, and be wise enough to know that you will want to change more than you think at first.As a very broad rule of thumb, niche software is priced as a function of what it would cost to write the damn thing from scratch. What you are buying is not a big saving in cost, but (usually) a dramatic decrease in risk. Bought in components have a fixed price and time, ones you build yourself do not.You need to do some rough bounds on what the system can do. Some systems by their nature cannot give competitve edge. Others only get noticed when they go wrong. You need to identify a few paths and model what happens at (say) 1%, 10% and 100% differences to default behaviour.Thus a pricing system that runs 1% faster may be worth a million bucks more, or nothing more. What is 99% availability worth vs 99.9% ?What is the mean time to repair vs failure time ? Excel apps crash with depressing frequency, but they are usually easy to fix. OK for many pricing models, but you don't want your settlements to go into a loop of sending money to people. (saw it happen once).If you replace 75% of someone's job, can you sack them ? If you can't move the other 25% to another person, you have a bored member of staff, and an IT bill both to build and run the thing. Easy for this to end up costing more.If you can't pose and answer these questions for your system then you are nowhere near being able to make a proper decision anyway.A big problem is personnel risk.I did a piece of really, really horrible code a few years back. Got into all sorts of places. Really ghastly it was. The comment in the code is "you won't understand this, ring Dominic on XXXXXXXXX". Alas, BT have changed my telephone number This sounds like a risk of building it yourself. In this case not so, I did it for a software house who sold it to a much larger firm who then sold it all over the place. All the major (and minor) vendors do this.

Thanks for your help all.Building my own is limited by resources at the mo. I have had a lot of suggestions as to what to avoid and, of course, it is very much appreciated. Are there any recommendations for off the shelf systems that are being used beside Algorythmics? What about Sophis?

Algo and Sophis target different market niches. Algo is for rapid calculation of VAR type risk. Sophis is good for the intra day management and hedging of risk in real time. Sorry - I assumed you meant the former - perhaps you mean the latter.I've used Sophis a couple of times now and think that for a small firm with limited resources it is a very appealing offering. The technical architecdture is simple, it has a reasonably robust and flexible data model and the model integration is not that hard. Speaking with them it look like they are keen to diversify their customer bas and push into the hedge fund market. Nice GUI and good docuemntation. I'm not sure it scales well (more than a few tens of traders, more than one geographical location) but for small scale it's good enough. For me the BIG plus of Sophis over Murex (which I would agree with the other poster - avoid) is thwe data model is very cclean and well docuimented - they seem to have taken a relational approach from the start.The other product you should maybe look at is Imagine - (their site is http://www.derivatives.com). They seem to be making most of their money acting as an ASP (app Service Provider) to hedge funds now. This is a smart idea because it virtually eliminates your set up cost (though you can deploy in house if you like). I was not so keen on their data model and their GUI is weak but they have some very strong yeild curve modelling facilities and are capable of better scalability (multi entity, multi location etc) than Sophis.I guess any converstaion like this has to include Summit - which depending on your portfolio may be able to do what you want. But integration with Summit canm be really tricky if you have not done it before.(all opinions are my entirely my own)