www.disktrading.com
$20 for what you require, including also I think per tick, 5m, 15m, 30m, 1h, and 1d data.
I bought some the other month, very helpful people.
$20 for what you require, including also I think per tick, 5m, 15m, 30m, 1h, and 1d data.
I bought some the other month, very helpful people.
singh2jatt
Member
- Messages
- 53
- Likes
- 0
HI i am after the same type of data. just wanted to know has anyone esle been susseful?
charting company tend to have this type of data is any1 good freinds with any1 in a charting company, or does any one know how i can get charts with this data. i would realy like this in a csv type format, as would be the best to back track, trading strageys without having to sit there for hours.
charting company tend to have this type of data is any1 good freinds with any1 in a charting company, or does any one know how i can get charts with this data. i would realy like this in a csv type format, as would be the best to back track, trading strageys without having to sit there for hours.
I bought quite a lot of data from disktrading (see above) for a project earlier this year.
The service is very efficient and cost-effective.
However the quality of the data (this was for the LIFFE Z contract) was not good enough for back-testing. I remember finding a lot of faulty data - it seemed like the contracts rolled over very untidily, as I recall.
The service is very efficient and cost-effective.
However the quality of the data (this was for the LIFFE Z contract) was not good enough for back-testing. I remember finding a lot of faulty data - it seemed like the contracts rolled over very untidily, as I recall.
I also used the Z data. When I imported in into NT I found that there were instances of duplicate rows pertaining to the first minute of the day on a handful of days, which I just removed.
For 1 minute data there are over 800,000 rows so there would have to be hundreds, if not thousands of incorrect rows for it to have a significant impact on a backtest.
Can you be more specific about he problems you saw?
For 1 minute data there are over 800,000 rows so there would have to be hundreds, if not thousands of incorrect rows for it to have a significant impact on a backtest.
Can you be more specific about he problems you saw?
I was experimenting with tick data in a table over 1GB in size.
There were frequent occurrences of series of ticks several points off the correct price. Additionally I remember seeing apparent bifurcations of the price (wild oscillations between two prices on alternating ticks), which I presume to be due to the prices for two successive contracts appearing together in the data.
The volume of data was far too great to cleanse manually and the nature of the aberrations was difficult to code around. It became rather apparent that there was layer-upon-layer of faults in the data - trying to fix it was like battling the hydra...
Trading signals are highly sensitive to such faults in the test data and the erroneous trades they generate are sufficient to have a very major impact on your results.
There were frequent occurrences of series of ticks several points off the correct price. Additionally I remember seeing apparent bifurcations of the price (wild oscillations between two prices on alternating ticks), which I presume to be due to the prices for two successive contracts appearing together in the data.
The volume of data was far too great to cleanse manually and the nature of the aberrations was difficult to code around. It became rather apparent that there was layer-upon-layer of faults in the data - trying to fix it was like battling the hydra...
Trading signals are highly sensitive to such faults in the test data and the erroneous trades they generate are sufficient to have a very major impact on your results.
Windowsill
Well-known member
- Messages
- 305
- Likes
- 39
10 years of 1 minute data!!!!! may as well take a week and just multiply it up 500 times...it wont look much different!
Which week? Last week!?
How can you possibly have any confidence to forward test a system if you have not back-tested against a variety of market conditions?
