How to enable "offline" Backtest
How to enable "offline" Backtest
25 Jan 2017, 19:59
Hi all,
real "offline" Backtesting, without active Sportware or intern connection, would be a great feature which I believe lots of people would appreciate.
So if you do please upvote the feature-request here:
Cheers
Mirko
Replies
ClickAlgo
28 Mar 2017, 16:59
See if this tool helps, it was built by Scyware for cTrader back-testing.
https://www.scyware.com/product/backtest-visualizer
@ClickAlgo
Mikro
28 Mar 2017, 17:43
RE:
Hi Paul,
this looks very nice. I'm just in the process of writing a csv export class to be able to export debugging and futher data to to some number crunching. This tool looks like it would come in very handy to see what exactly is happening in the bot. Being able to plot drawing objects in the backtest, like writen in the planned features section, would be a great help, too!
I'll look into this asap.
Thank you!
Paul_Hayes said:
See if this tool helps, it was built by Scyware for cTrader back-testing.
@Mikro
Mikro
31 Mar 2017, 23:16
Scyware Backtest Analyser
Hi Paul,
Do you have any Contacts to Scyware? I'm asking because I find the BacktestVisualizer pretty helpful but I would like to extend the analysed Data to monitor internal Data in my Bot.
Or following another approach, have you made experiences with MATLAB and cTrader?
Writing a .csv from within the bot an then analysing it is nice. Really usefull would be writing them to MATLAB during Runtime and perhaps retourning processed Data.
Cheers
Mirko
@Mikro
ClickAlgo
31 Mar 2017, 23:31
Hi Mirko,
I do not sorry, also consider looking at F# and R.
@ClickAlgo
gonzsanc
12 Oct 2019, 20:28
RE: Scyware Backtest Analyser
Mikro said:
Hi Paul,
Do you have any Contacts to Scyware? I'm asking because I find the BacktestVisualizer pretty helpful but I would like to extend the analysed Data to monitor internal Data in my Bot.
Or following another approach, have you made experiences with MATLAB and cTrader?
Writing a .csv from within the bot an then analysing it is nice. Really usefull would be writing them to MATLAB during Runtime and perhaps retourning processed Data.
Cheers
Mirko
Just to mention that Python (native) has a lot of data analysing extensions. If you use cython or mumba, you can get near-to-c++ speed in your code. Otherwise, it is quite slow.
You basically can use Numpy for fast analysis and Pandas for matrix management.
R is also good, but it is a limited language since it is oriented for statisticians, not for development as Python is.
Then, for advanced statistics, Python is really excellent with packages such as sklearn, Tensorflow or scipy.
The advantage of .Net is that creating GUIs with Python is not really a pleasant task.
However, for most simple maths and stats, .Net's LINQ is enough.
@gonzsanc
Jimmy
26 Jan 2017, 22:24
Offline backtest
Yes this is a primary need, it HAS to be possible to do offline backtesting with imported data.
I have written some highly profitable robots in MQL but i do not trust them, i know what 'they' can do with my trades on the other side.
Somehow i feel that there exists an element of trust when it comes to spotware, don't ask me why.
Now my idea is to convert my MQL robots to Calgo but if i can not run a backtest offline i will not proceed.
I have historic data from 1970 up to today.
I see NO reason as to why the terminal has to be online for me to run these backtests.
Please implement the feauture.
@Jimmy