Tag Archives: Theory

Yes another book on Professional Automated Trading: Theory and Practice

Yes another book on Professional Automated Trading: Theory and Practice

This came in yet again from the NYC contact

 

http://books.google.ca/books?id=xWo5AQAAQBAJ&pg=PT11&lpg=PT11&dq=Professional+Automated+Trading:+Theory+and+Practice&source=bl&ots=OkUf-MVko4&sig=TYgqhP8dTrlCu3oWDgkbXxYECG0&hl=en&sa=X&ei=Q84OU8LuDezJ0gGgmICwBg&redir_esc=y#v=onepage&q=Professional%20Automated%20Trading%3A%20Theory%20and%20Practice&f=false
Join my FREE newsletter to find out which books I refer to for automated trading

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

Learn Random walk theory for market inefficiency

Learn Random walk theory for market inefficiency
Get more details here to get access to the course and other huge benefits including High Frequency Trading platform building, QuantLibXL analysis, Matlab, etc. http://quantlabs.net/quant-member-benefits/slash-your-quant-learning-curve/
Thanks for reading Bryan

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

Future models, algos, and strategies to be delivered for matlab r c++ C# c sharp and latex with video theory lesson

 

Future models, algos, and strategies to be delivered for matlab r c++ C# c sharp and latex with video theory lesson

Get notified when this is ready! Be notified here.

 

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

The effects of Black Swan Theory on quant advanced neural networks

The effects of Black Swan Theory on quant advanced neural networks
This was a discussion found on Linked In:
The technology has progressed since the 70’s. A follow-up would be have you recently used a NN to predict something about a financial market?

Either way is fine and there is nothing wrong with knowing other peoples opinions but sometimes there is no substitute for actual doing or even just trying.

I’ve read the Black Swan book. From the limited view of traditional statistics he’s academically correct: something unexpected can happen. However, overall I found his perspective totally irrelevant to what I do. One is not trying to predict very event, every moment in time. If you are then of course what do you do about that rare Black Swan moment?

In trading the only requirement is that on those specific moments when your model makes a prediction that it is accurate most of the time. As the model is developed from history and as Black Swans are so rare they are eliminated as outliers in preprocessing, the NN will not see them and so not act on them.

Is your perspective that the market is fractal so it can’t be predicted accurately?

If it can be predicted does that mean it is not fractal or that the NN is capable of prediction for fractal functions, strange attractors, etc.?

None of our models will pick up the Black Swan type events by definition. All we can do is try to hedge/insure our positions against these events when economically sensible, which is Taleb’s philosophy. Buying far out of the money puts on the stock market when you are highly leveraged long due to your model’s prediction is a perfect example.

Also there there are smaller day to day signals that our none of our models will pick up that we can call “noise”. For example if new information becomes available that we just haven’t captured in our inputs, it will affect prices.

So the best we can do is have more/bigger winners than losers.

It is proven that a neural network can approximate any functional mapping to any desired degree of accuracy. So if the information is in the inputs, I know that a properly constructed net or committee of nets can tease it out and make money.

My philosophy has been that the value of market information decays quickly and that predictions made more for more than the very short term are suspect. But I see in some of the comments that people are predicting much farther into the future. I am curious to hear about the effectiveness of these approaches.
Actually, a model which picks out the black swans is a rather dubious one, i guess. Even if news impact is moving towards higher and higher frequencies, you can still catch it, based on a (often nonlinear) reinforcment loop/feedback.

NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!