move into trading?

[bored]

The Mac kernel that underpins Mac OS X for example automatically multitasks Java threads over multiple CPUs on multi core Macs which max out at 8 cores of 3 Ghz each IIRC.

Just FYI, any modern OS will multitask any program over multiple CPUs/cores unless you specify otherwise. This is in no way Java or Mac specific (and is not the reason why Java is the language of choice).

[HarryPearce]

True, but OS X is based on BSD and maps Java threads to pthreads which are managed by the OS rather than the JRE. Neat if you ask me since Java's libraries seem easier to understand and use than native code libraries (Unix or Windows). Its not the best multi tasking system I've seen; that compliment belongs to Termite http://p-cos.net/lisp-ecoop05/pdf/19654.pdf which is similar to Erlang but written in Scheme. Sadly, Gambit which underpins Termite doesn't provide support for multi-cored CPUs even though it can run millions of concurrent threads.

Nick

[Buffoon]

OK, hands up time. Yes, I moved from development to trading and then on to my own business. It does happen. Algorithmic trading is one of the things that happen here.

Almost all stuff is written in C++ and C# (with a tiny bit of VB). There isn’t a MAC in the place, yes there are Linux boxes here and there are some java applications but they are few and far between.

I’ve interviewed candidates who spout the java/unix extrament for automated trading; they are pompous arses who have never held a position in their lives and no nothing about trading. I wouldn’t give them the time of day anymore. As soon as they say the word ‘latency’ have them thrown out.

[wendigo100]

Yeah, me too.

[brownie74]

I still remember my first day contracting in the City in '94. I was shown round the desks said hello etc etc and then came across one desk with no-one sat at it, but the books said enough a guide to Standard ML and another text on Lattice Theory (If you need to ask waht they are you don't need to know!).
Nick

I worked with ML/Scheme compilers in 97-98 for a big fin services company in the square mile. They were doing calc's over huge list of numbers (historic share prices). Writing list processing code is quite nice in these syntaxes. I remember trying to overload an operator in ML and the compiler barfed out an error message of "you have got to be kidding pal".

[GreenerGrass]

OK, hands up time. Yes, I moved from development to trading and then on to my own business. It does happen.

If you don't mind me asking, how did you make the jump? To go from experienced IT contractor to novice/trainee trader would you take a big initial drop in income (to hopefully be ramped up again later plus bonuses etc.)?
I guess another factor is whether you want to be tied to London even more (which I'm guessing you would be as a trader...)

[brownie74]

also, you'd be entering the trading world at the age most of the traders are leaving it. early 30's, no? do they care what age you are? or does it mainly attract younger people because of the long hours / fastlane lifestyle?

[BlasterBates]

With algorithmic trading the understanding of how it works is really important, so I think there would be opportunities. If you're new to trading then you need to learn finance first.

[Troll]

That you'd be moving into trading implies the trading would be fully or near fully automated and you would be expected to develop the trading models.

Nick

Nick... just reading that gave me a headache! do you actually enjoy & get personal satisfaction working at this level? if so what are the highs & lows?

[BlasterBates]

I suspect there's some element of truth in the idea of moving into trading.

I used to work in AI within the City and some of the work I saw was well into the realms of rocket science requiring supercomputing of the Cray or near Cray standard. My MSc examined a good deal of the financial trading systems that were about around ten years ago, and considering the advances in AI and computing power I'd be very surprised if the quants desks were not heavily involved in certain areas of the crossover between computational learning, stats, maths and high powered computing platforms.

I still remember my first day contracting in the City in '94. I was shown round the desks said hello etc etc and then came across one desk with no-one sat at it, but the books said enough a guide to Standard ML and another text on Lattice Theory (If you need to ask waht they are you don't need to know!).

As to the requirement for Java and multithreading I can only hestitate that they're working on heavily distributed systems for which Java excels. The Mac kernel that underpins Mac OS X for example automatically multitasks Java threads over multiple CPUs on multi core Macs which max out at 8 cores of 3 Ghz each IIRC. To get max performance out of such would require PhD level knowledge. Never mind the need for similar maths knowledge to hande the quants side of things.

That you'd be moving into trading implies the trading would be fully or near fully automated and you would be expected to develop the trading models.

Nick

Was this the era of neural networks?

That's all gone, algorithmic trading is by comparison quite simple.