Someone sent me questions on this requestting to see more:
> I need more info
> – This system/model works on Forex and MT4 ?
–> Metatrader only limits you to one data connection and execution. These are tied into probably a broker you don’t even know that is trading against you see below. The data could be questioned as well. See below. Platforms like Ninja and Metatrader are severely limited. They are called black boxes which I don’t like or support. I only use open source trading platforms which my membership knows about.
> – Wich broker are you recommended?
–> One that I like is Interactive Brokers and a few others. Others I cannot mention are corrupt. They will send out trades to work against you. These are called bucket shops. If I mention them by name, I ame sure I would get some lawyer after me. Just do your research! http://www.investopedia.com/terms/b/bucketshop.asp
> – Is this type or trading allowing by brokers?
–> Is your broker registered? If not, find one as they have to play by the rules.
> – Are the some specific system working already?
–> Sure I am working on that
> – How much profit in pips can I get ?
–> I am not sure.
> Thanks in advance..
Ensuring homogeneity and heterogeneity within modelling segments for quant analytics uses
Can anyone suggest a good statistical technique to ensure that the modelling segments are heterogeneous but the segment itself is homogeneous??
It seems like the major task of clustering , so search for measure that estimates quality of clustering structure
Agree for an unstructured approach. If you have a known outcome variable you want to segment on, try decision trees.
ANOVA would be another approach – if as John Parker (above) says, you have a known outcome variable and it is continuous (interval or ratio). If categorical, a Contingency table should allow you to visually see where the majority of cases lie for each of the groups.
If I am understanding the question correctly, there is a problem with ANOVA and similar approaches in that the task of clustering is to maximise differences between segments and so asking whether they are then significantly different breaks the rules. Though indices such as F can be used as a qualitative way of seeing how differentiated each segment is from others.
To my mind, looking for segment differences on an eternal variable (that is, one not used in the clustering) is a helpful way of seeing whether a particular clustering solution is meaningful. I am also assuming that the clustering has been repeated on split samples to show that the same solution is appears from sample to sample (and so is less likely to be a solution that capitalised on chance).
Computer quant anlytics modelling: Brain in a box with nature news and comment
Henry Markram wants €1 billion to model the entire human brain. Sceptics don’t think he should
I tend to view computer modeling as a bridge between theory and experiment. In the case of the brain, theories are a bit lacking. I would love to hear what Jeff Hawkins has to say about the HBP since he built an institute to create real theories of the brain.
Yes, “visionaries are sometimes necessary to drive progress.
This Matlab webinar is the heart of my quant analytics for strategy development and modelling a profitable algo
This is the Matlab webinar that will start getting me onto my strategy development. Cointegration is pretty big which includes a lot of the basics of strats used in HFT environments. I thought this would be a good place to start with others as well.
Quant analytics: Profitable Trading Strategy development, algo and modelling Meetup is now created
Profitable Trading Strategy development, algo and modelling Meetup is now created! Please join for updates on future meetups both online and face to face. This will be identical to meetup.com/quant-finance as well.