Best tools for statistical analysis of tick data and news events? | Forex Factory

Options

Permalinks
Print Thread

Similar Threads

What platform has tick by tick (multiple chart filling) data replay? 4 replies

Tick Size, Tick Rate, Tick Spread, and Tick Logger indicators 36 replies

Where can I find tick-by-tick data for various currency pairs? 9 replies

download data tick by tick ??? help 1 reply

Best tools for statistical analysis of tick data and news events?

  • #1
  • First Post: Apr 24, 2012 12:20pm Apr 24, 2012 12:20pm
hey folks,

I was wondering what the best tools are for statistical analysis of tick data... and news events.

I guess the best tool for testing trading systems would be mt4/5, code the system, do a backtest with high quality history data...

but what about statistical analysis of crossrate data? like, for starters average winning days in a row for example to give an easy example...?

Or average daily range on some news?

I was thinking about coding that stuff in java, but maybe there are easier ways to do statistical analysis of forex data... mathlab? statistica?

would be great if anyone could point me into the right direction.
  • #2
  • Apr 24, 2012 4:36pm Apr 24, 2012 4:36pm
Quoting herbun
hey folks,

I was wondering what the best tools are for statistical analysis of tick data... and news events.

I guess the best tool for testing trading systems would be mt4/5, code the system, do a backtest with high quality history data...

but what about statistical analysis of crossrate data? like, for starters average winning days in a row for example to give an easy example...?

Or average daily range on some news?

I was thinking about coding that stuff in java, but maybe there are easier ways to do statistical analysis of forex...
JavaScript is the probably the easiest way, but you can do it in any language you're familiar with.
  • #3
  • Apr 25, 2012 4:17am Apr 25, 2012 4:17am
In my opinion javascript would be amongst one of the worst languages for the job... one of the reasons why is that you are limited to a browser doing the work, which considerably slows down the process, maybe even stability...




I've been researching a bit and R seems to be a great environment, because it is similar to S and S Plus, and a lot of literature of statistical analysis of financial markets / time series uses these platforms for analysis. there are even special packets for financial analysis for R, such as quantmod.


So I guess I'll go with that...
  • #4
  • Apr 25, 2012 7:28am Apr 25, 2012 7:28am
Quoting herbun
hey folks,

I was wondering what the best tools are for statistical analysis of tick data...

I was thinking about coding that stuff in java, but maybe there are easier ways to do statistical analysis of forex data... mathlab? statistica?

would be great if anyone could point me into the right direction.
Hi, take a look at R.

http://cran.r-project.org/

Here you can find a lot of packages:

http://cran.r-project.org/web/packages/

I think R is a good free statistical tool.

Have fun

Montecristo
  • #5
  • Apr 25, 2012 10:20am Apr 25, 2012 10:20am
I've been looking into that, but I was not able to even find a solution to a normally very simple question:
how do I count the days in a row a timeseries is short or long?
like for example...
99 days it was 1 day in a row long,
30 days it was 2 days in a row long etc...
once thats done, plotting it to a graph or applying some indicators to it won't be that hard, most tutorials concentrate on that... but I have no clue how to do the first part...?

Or is R not the right software to use for the first part, and I should manipulate data in another language like java, and then do statistical analysis on the results in R?
  • #6
  • Apr 25, 2012 12:44pm Apr 25, 2012 12:44pm
my understanding now is that R is for data analysis, while the example I was talking about that causes me trouble (number of long days in a row) is actually part of specific data collection from raw data, not the analysis itself... so to speak.

so guys... any software to help with that? or does that have to be coded in whatever language I prefer?
  • #7
  • Apr 25, 2012 4:46pm Apr 25, 2012 4:46pm
Quoting herbun
In my opinion javascript would be amongst one of the worst languages for the job... one of the reasons why is that you are limited to a browser doing the work, which considerably slows down the process, maybe even stability...




I've been researching a bit and R seems to be a great environment, because it is similar to S and S Plus, and a lot of literature of statistical analysis of financial markets / time series uses these platforms for analysis. there are even special packets for financial analysis for R, such as quantmod.


So...
JavaScript beats R in performance for a lot of things. It's too bad people only associate it with web design.

http://code.google.com/p/v8/

In the past I've used this library:

http://www.jstat.org/

It's worth using if you don't want to learn to use something retarded like R.
  • #8
  • Apr 26, 2012 4:30am Apr 26, 2012 4:30am
LOL. A javascript standalone engine. Did not know about that one, thanks.

The more I learn about R, the less I like it, too... The programming is really retarded for someone who's been used to program in other languages...
plus, its not a performance champ either...

And yesterday I learned that after all, the main problem is not really doing statistics calculations, but to prepare the data first... and R won't be of any help with this. :/

I'm still reluctant to use javascript, because I can only do basic stuff with it, but then again, it might gain additional benefits if I learn the language. hmm.

Thanks again for the links, I'll take a closer look for sure.
  • #9
  • Last Post: Jul 4, 2013 9:56pm Jul 4, 2013 9:56pm
I see this is an old post, but I just got annoyed by reading this. If you do not know how to use R, maybe blame yourself first. Usually it is high end language, which lets you easily manipulate the data and combine functional programming with a bit perverse oop. However perverse is not necessarily in a bad way. Furthermore you can choose 1 out of 3 different class definition approaches or combine them all. Anyways, draft in R, once you're done, write your functions in C or Cpp. This way you have efficient computations and have the newest statistical analysis tools. Alternative could be python + numpy, scipy etc. However R is generally preferred due to its community and variety of libraries. One of disadvantages for someone who cant do C, Cpp or fortran is speed. Another is, if you are programmer and you want to stick to C family go python, however I would diversify your skills. Even if I know python and matlab I go for R due to its nice and easy syntax. with 2 lines you can do more than 10 lines in alternative languages.(exaggerating...)
0 traders viewing now