Linear SVM's for daily range estimation using Rapidminer - page 3
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Hello Mr Tools and Alex,
Would you please be so kind and also attach the indicators you use when referring to them since Forex TSD doesnt make it too easy to find the relevant ones.
Thank you very much and have a good weekend
Daniel
Hello Mr Tools and Alex,
Would you please be so kind and also attach the indicators you use when referring to them since Forex TSD doesnt make it too easy to find the relevant ones.
Thank you very much and have a good weekend
DanielHi Daniel,
The Wolf Wave I just posted here
https://www.mql5.com/en/forum/180648
the last post on that page, and i am just experimenting with it, but the 5 so far has stopped sliding around 5 pips from the projected support line, if i remember right think Hurst projections he said to allow for 10% difference, so haven't done the math so don't know, but 5 pips is darn close.
Hi Daniel,
I am using this one....
https://c.mql5.com/forextsd/forum/91/alb_mogalef_bands_amp_projection_no_sunday.mq4
A quick note on inputs
For a technical overview of SVM kernels:--
Range
Storing the data to the repositories solves many issues with Rapidminer when it comes to how Rapidminer deals with metadata. The main building blocks of the script are outlined below but there are quite a few things to explain. Scripts can be nested to run in parallel and in sequence to make best use of multiprocessor machines. This script uses two linear svms for the highs and lows but they are isolated in the sense that the low values are excluded for the the svm that is looking for patterns in high values. What affects the end result are the inputs and how the values of "C" are optimised for the svm. This will take more time to explain and it dependent on the time series you are looking at. The number of decimal places for example can change the way one optimises for "C".
Alex,
Congratulations for your excellent thread.
I have a suggestion,in case you find it interesting:Add a third linear SVM trained on range and see if this add something or not.For example,you may find out that in instances when you get,for example,high=1.32000,low=1.31000 and range =0.01+-,the precision of the estimated high-low levels is higher than in case you get a "non matching" range like 0.005 or 0.02.
S
Hi Simba,
I will give it a try. I am also looking at range in terms of cycles and standard deviation. I have one script ready to go with three svm's so I can test your idea quite quickly. The third SVM was trying to predict closing prices, actual close and also close location value as range between -1 and 1. Either way predicting close was a lost cause at least with the inputs I was using so I went back to two.
I am now looking at turning points in terms of standard deviation and mladen has made a tool that is ideal for that. The idea is that if we have a certain info about the ranges in terms of standard deviation then we can see if a predicted high or low is exceeding that threshold and by how much. It could hopefully give a one day prediction on when trends might end.
Here is a snapshot from a few moments ago which helps visualise the idea.
regards,
Alex
While this is on my mind, here are the pivots for today.
and a pretty decent estimation so far for today...
BOJ easing- Much to do about nothing in terms of range.
Projected Pivots and Options
Here is something I have been following for several days and many thanks to Daniel for pointing me in this direction. There is a very strong correlation between CME option levels and the projected pivots. The SVM's seem to be predicting with very high accuracy the levels where open interest is high. This correlation has been present daily over the last week of trading.
Attached is a picture of today, where the SVM predicted low has a 100% match. This probably goes a long way to explain how this type of black box can have any prediction value at all in markets that are essentially news driven.
To conclude, SVM's can be a useful tool for analysing forex time series but with a high costs in terms of time and computing resources. At this point, I won't be adding anything else to this thread as the building blocks for this kind of system are outlined for anyone who wishes to explore what is possible.