Okay first of all, this data is free so this is not a complaint or rant thread.
I have just recently got into programming, and have been doing lots of googling to try and find the best data to use for backtests. Apparently Alpari M1 data is supposed to be good, or possible Dukas data too, but I have been unable to find those in searching, only found old dead links on some sharing sites, so I am not able to compare. If you have compared more than one M1 data source, please share your thoughts. Also if you know where and how to obtain this data please share. I did find one other source for M1, Hisdata.com has Tick AND M1, no idea how good either is?
See my charts below, on why I think the Forexite data doesn't look so hot. Now these are M1 charts, and really for M1 you would want Tick anyway, but using this kind of M1 expanding onto M5+ charts, makes me wonder about how good M5+ is really going to be, surely there is better M1 data out there? I looked through a bunch of this Forexite data and unfortunately it all looks like this. Notice in the charts that this is exactly the same chart/time period etc, only one picture was taken on wide screen monitor and the other not. (also one is the GMT 0 and the other is not [+3 looks like] but I was VERY careful to take a pictures at the exact same time in past.
First chart is AJ M1 Sep 26 Forexite ~9pm GMT 0, 2nd chart is same equivalent but Pepper data. Look at the difference in quality of candles, eegads , and even the overall shape has some significant differences here. 3rd chart is Forexite, and 4th Pepper equivalent, again quality of data is no comparison to me, at least the path of price is more comparable here (usually the case).
Now I can say that I have looked at M5, and M15+ with these same two data sources (Pepper and Forexite) and it does "appear" to NOT have to big effect overall, on the big scheme of things, but this is only my inexperienced off the cuff opinion, again please chime in those with more knowledge here. The overall shape of things even on M1 is more or less the same, and the higher you go the more it seems to take a more accurate shape.
Of course, anyone who wants to test scalping strats needs, Tick data to try and give some kind of accurate representation of the past and achieve the best possible modelling quality of 99% is the only way to go (or better yet don't use MT4, if you can). For others testing out possibly M5, or especially M15+, using M1 data and getting 90% will be likely good enough overall in most cases (from what I have gathered), at least to get the ball rolling for sure. The trouble with the Tick is they are very LARGE files, and downloading and importing is major PITA. (see Birt's site)
TIA for any insight here
p.s One other question about Tick if anyone knows, if I have Winblows (hehe) on a 30 gig partition and need MT4 to work with all these Tick files, which I heard can add up to 100 gig, will I need to merge this partition with another to make it bigger?
I have just recently got into programming, and have been doing lots of googling to try and find the best data to use for backtests. Apparently Alpari M1 data is supposed to be good, or possible Dukas data too, but I have been unable to find those in searching, only found old dead links on some sharing sites, so I am not able to compare. If you have compared more than one M1 data source, please share your thoughts. Also if you know where and how to obtain this data please share. I did find one other source for M1, Hisdata.com has Tick AND M1, no idea how good either is?
See my charts below, on why I think the Forexite data doesn't look so hot. Now these are M1 charts, and really for M1 you would want Tick anyway, but using this kind of M1 expanding onto M5+ charts, makes me wonder about how good M5+ is really going to be, surely there is better M1 data out there? I looked through a bunch of this Forexite data and unfortunately it all looks like this. Notice in the charts that this is exactly the same chart/time period etc, only one picture was taken on wide screen monitor and the other not. (also one is the GMT 0 and the other is not [+3 looks like] but I was VERY careful to take a pictures at the exact same time in past.
First chart is AJ M1 Sep 26 Forexite ~9pm GMT 0, 2nd chart is same equivalent but Pepper data. Look at the difference in quality of candles, eegads , and even the overall shape has some significant differences here. 3rd chart is Forexite, and 4th Pepper equivalent, again quality of data is no comparison to me, at least the path of price is more comparable here (usually the case).
Now I can say that I have looked at M5, and M15+ with these same two data sources (Pepper and Forexite) and it does "appear" to NOT have to big effect overall, on the big scheme of things, but this is only my inexperienced off the cuff opinion, again please chime in those with more knowledge here. The overall shape of things even on M1 is more or less the same, and the higher you go the more it seems to take a more accurate shape.
Of course, anyone who wants to test scalping strats needs, Tick data to try and give some kind of accurate representation of the past and achieve the best possible modelling quality of 99% is the only way to go (or better yet don't use MT4, if you can). For others testing out possibly M5, or especially M15+, using M1 data and getting 90% will be likely good enough overall in most cases (from what I have gathered), at least to get the ball rolling for sure. The trouble with the Tick is they are very LARGE files, and downloading and importing is major PITA. (see Birt's site)
TIA for any insight here
p.s One other question about Tick if anyone knows, if I have Winblows (hehe) on a 30 gig partition and need MT4 to work with all these Tick files, which I heard can add up to 100 gig, will I need to merge this partition with another to make it bigger?