Hi,I am currently applying for an artificial intelligence program in VU in Amsterdam. As I am already working in finance and would continue to work in finance after this graduate degree it is important to me that the education can be related to finance. The current list of specialisations Click hereIt goes without saying that if some specialization is totally uninteresting or not really applicable to anything practical I would like to know this also. By previous background is a five year program (BSc & MSc) in business & economics (major in Financial Economics). Thank you all in advance for your insight and helpBest Martin

Hi Martin,I am doing a PhD. on Machine Learning Techniques in Finance, in Intelligent Systems Lab, UVA, Amsterdam. I have been struggling to find a mid-point between finance and Machine learning as well. I have to admit it's not an easy challenge to satisfy both fields.The subject specialization mostly depends on your interests and your abilities. But from what I see on VU's page, Technical Artificial Intelligence seems to be the most suited one, if you want to apply the techniques directly on financial markets. This includes automated agents, pricing, modeling, forecasting(regression) etc.I guess if you'd like to stay as an academic, Computational Intelligence and Self-organization is also a very interesting topic. I think nowadays, behavioural economics is quite popular in academia, rather than the regression kind of stuff. I believe everyone in academia is kind of aware about the fallacy of the models, which is actually true for any field.I am working in WTC, if you'd like a chat in person, send me a PM.Cheers,

just speculating: maybe communication, as in interpretation of news releases followed up by adjusting positions..

QuoteOriginally posted by: taylanHi Martin,I am doing a PhD. on Machine Learning Techniques in Finance, in Intelligent Systems Lab, UVA, Amsterdam. I have been struggling to find a mid-point between finance and Machine learning as well. I have to admit it's not an easy challenge to satisfy both fields.The subject specialization mostly depends on your interests and your abilities. But from what I see on VU's page, Technical Artificial Intelligence seems to be the most suited one, if you want to apply the techniques directly on financial markets. This includes automated agents, pricing, modeling, forecasting(regression) etc.I guess if you'd like to stay as an academic, Computational Intelligence and Self-organization is also a very interesting topic. I think nowadays, behavioural economics is quite popular in academia, rather than the regression kind of stuff. I believe everyone in academia is kind of aware about the fallacy of the models, which is actually true for any field.I am working in WTC, if you'd like a chat in person, send me a PM.Cheers,Thanks, I think such a conversation would be very informative for me Sadly, I currently work in a Research Institute on the French Riviera so its quite hard for me to get to Amsterdam. However I might be going to Amsterdam soon, so I might take you up on your offer in the near future. Regarding the Technical AI; it kind of scares me that I do not have much formal training in computer science. Pursuing a specialisation which clearly states that it is made for CS majors seems risky as I dont know how they shape the program. If they build it directly on prior courses it is clear that I would have some problems as I never had the foundation courses. So maybe I should call someone there so ask what they really mean with their formulation. Your second suggestion seems very interesting! Looking through the list of courses it seems that this could be highly educational for me! Ill definitely investigate this option further. Thanks for helping me narrowing it down a bit. But what about you? Do you feel that AI is applicable to Finance? Do you feel the relation between finance and AI is growing stronger or weaker? Is there any academic interest for AI in finance or interest for finance in AI?

AI is indeed applicable to finance, but I'm not sure that it is applicable by you.You say you don't have a background in CS and Bus/Econ seems to me pretty much the subject I'd least choose if I was going to do AI in a bank.Let me explain my harsh words...AI in a bank consists of doing things to computers that do things to data to make money.In other words they will expect you to write code in some mix of procedural and functional languages and with no real programming background I don't see how any course, no matter how good, can get you to the stage where you could do this.A growing area of AI in finance is making sense of the many text based knowledge sources on the web and other streams. This is often called sentiment analysis which sounds better than it is. Firstly I have to report that this niche seems to be full of charlatans and/or those overworking tools that are not up to the job and a particular issue is that they are struggling to get much beyond "people like/dislike this"That's because it is a hard problem and there are some very sharp people trying to solve it.The reason I focus on this, is that I'm trying to work out why you think this course is good for you and I suspect that like others you see an opportunity for yourself there.I don't.You don't mention any background that would put you ahead of others.It's actually quite rare that I directly say to someone "don't do this course" but unless there is some major part of your skills and experience that you've not shared I don't see this helping your career at all.Conversely I'm not saying it is a bad course, indeed if it's good, then you would not be able to do it since some lectures should start with:"Let's revise what we know about lambda functions...""Can we use blackboard systems in understanding free text ?""Is the set of functions computable by deterministic Turing machines less than the set computable with non-determinism ?"Please tell me if any of those intros represent things you know already.

QuoteA growing area of AI in finance is making sense of the many text based knowledge sources on the web and other streams. This is often called sentiment analysis which sounds better than it is. Firstly I have to report that this niche seems to be full of charlatans and/or those overworking tools that are not up to the job and a particular issue is that they are struggling to get much beDo you know any techniques/products which do it based on Linguistic techniques (OpenAmplify a startup from Baltimore Maryland attempts this. It does it reasonably well depending on what you need). So much is done using some form of Classification Techniques - but it does not work very well as you cannot have enough training data to make it work. But Linguistic techniques have the inherent limitation that they products tend to be language specific. The other hard part about Sentiment Analysis or Opinion Mining as it is called is discovering sentiment/opinions by topic. Level of emotion is also key to understanding. Most tools based on classification methods will tag a sentiment at document level which if of limited use. But people express many ideas in one document and feel differently about each one. It is not easy to find good material on these topics. Can you suggest anything that you may have heard of.Do you know if any research and good reading material in this area. Books vary wildly between discussing arcane statistical techniques or demonstrate over-simplified nonsense like Bayes Classifier which in my opinion does not work well but even the uninitiated can understand.

I have to admit that it is quite a harsh reply to someone newly starting the field. I don't mean to offend you, but I don't agree with many points in your post. Being in the machine learning and pattern recognition field for some years, let me try to demystify some of the concepts in your reply, and at least try to encourage any new-comers to the field:- AI as it is, is more a hype word than realized intelligence. It is actually a hype word from 60s, and the current replacement is Machine Learning. The field brings together many fields, particularly statistics and probability theory, and can help to generate models for conceptualizing the real world. As models are ALWAYS wrong, It can only act as an assisting technology to making money, like analyzing market movements, finding patterns etc. Mostly however, people don't use AI to make money directly. They use simpler models to do that. - You will be expected to write programs, and be good at it. This is a fundamental reality of our age, and I know many traders that have started learning programming, and have become quite proficient in this. If you think about it after all, studying sth is actually for learning sth. If there is nothing that you will learn, and you do it only for the title, I think you rather do an MBA. But since you are curious about the topic, I believe you are ready to take the challenge.- As far as I know there is no single algorithm yet, that can extract 'tips' from text (i.e. internet) about what will happen to instruments. If this could be done, google would have been a finance company. There are however, expert expectations on many financial figures, and they publish their expectations on what will happen if the anounced figure is higher/lower than the expected. But this is not artificial intelligence, it is using human intelligence.- I believe anyone crossing the boundaries of fields can bring in some new insight possibly for both fields. If you're stuck learning one thing only, you will start looking everything from one single view point, and won't have any edge anyway.Apart from these points, my 'encouraging' questions would be:- Do you think huge amounts of numerical data can best be interpreted by computers?- Are you interested in dealing with this kind of data, and forming models, visualizing or extracting information from this data?- Are you interested in increasing the robustness of your models for unforeseen circumstances?- Do you think we are stuck in 3/4 dimensions, just because our senses tell us so?If yes, do you mind exploring possibilities for expanding your senses to multiple dimensions?

@capafan2 I agree that it's really tough to find actually useful sentiment analysis work@talyan, don't worry, I'm not offended, I wrote a Prolog in Lisp as an undergrad and have done various things with AI/ML over the years, dig up very old things about Fuzzy Logic, you might find my name (we all make mistakes, OK ?)So although you may think you're correcting some fat old headhunter, which of course I am, be aware that I've done AI for money before you try to "demistify" my posts.I was writing as a headhunter, not a computer scientist.A line that I push as hard as possible for a newbie to seek out the areas in which he excels and enjoys, martingull presented himself as someone with no apparent reason to believe he has aptitude and experience beyond that of a reasonably bright person who has studied a very different subject. I see the risk/return in this context as very poor.I tried to point out that I had no issue with this course, just it's suitability for him.You open an interesting line when you say "If this could be done, google would have been a finance company"They are doing this, UCEM is very clear on this, but it's not (yet) smart enough to work out what the fuck they are doing, it may be be fixed income but even that has a low confidence.As above I can't support your encouraging quesitons, it's not my job to encourage or discourage, it is to give the best advice I can.

I am working in Amsterdam in a trading position and just recently had a look at Russell/Norvig. I started reading it out of pure curiosity as I had absolutely no idea about the subject. It looks pretty interesting, not sure how applicable it is in my field, though. If you are back in Amsterdam and feel like having a chat, drop me message.

I would say that Machine learning is a subset of AI. (other areas are eg computer vision/reasoning/language)I don't believe sentiment analysis is currently being used much ( except to bring in investors!)I would say Machine learning & statistical models for huge data sets are being used.( don't think all the autonomous agents stuff is being used..)Its rather unclear which of computational intelligence or technical AI modules cover this.You might want to look at the nuclear phynance forum, which has more hedge fund quants than derivatives quants.wrt to machine learning - i can't see that a comp sci degree is "relevant" - arguably the main subject is statistics, so you might have the prerequisites from economicsmight want to look at online book to assess your suitabilityelements of stat learningrebonato has a book on bayesian networks: Coherent Stress Testing: A Bayesian ApproachCoherent stress testing - bayesian approach

QuoteOriginally posted by: spv205I would say that Machine learning is a subset of AI. (other areas are eg computer vision/reasoning/language)I don't believe sentiment analysis is currently being used much ( except to bring in investors!)I would say Machine learning & statistical models for huge data sets are being used.( don't think all the autonomous agents stuff is being used..)Its rather unclear which of computational intelligence or technical AI modules cover this.You might want to look at the nuclear phynance forum, which has more hedge fund quants than derivatives quants.wrt to machine learning - i can't see that a comp sci degree is "relevant" - arguably the main subject is statistics, so you might have the prerequisites from economicsmight want to look at online book to assess your suitabilityelements of stat learningrebonato has a book on bayesian networks: Coherent Stress Testing: A Bayesian ApproachCoherent stress testing - bayesian approachThanks for a great book and a seemingly great forum!

@celala That would be great. I'll PM you when I am in Amsterdam