Thanks Fan27.. can't wait for that course to start ! 
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
Targets on Machine Learning approach
- Thread starter HarrySquatter
- Start date
Hello!
I'm pplaying with different machine learning algorithms and stock data but I'm a bit confused. Can anyone tell me what is the target algorithm accuracy when it comes to predicting next movements (up/down) upon which one can build a strategy? Should it be 51% or 60, maybe sth different? I'm just trying to build my intuition when it comes recognizing the results that I'm getting from the models that I build. Any suggestions on this topic would be appreciated.
Sorry your thread got hijacked.
To answer your question: it depends.
It will depend, at least, on the size of your profits vs. the size of your losses.
Hallo,
I'm interestet in learning more about using tools/platforms or own programs to find edges in the market. Can anyone recommend how to start, recommend tools and/or simple sample programs/algos that help getting my toe into the cold water slowly and gain some experience ?
I have experiance in C/C++ and use Sierra Chart. It would also be interesting to know if it makses sense to use Sierra Chart ACSIL for that purpose or if it is necessary to write a compülete tool by my own.
Looking forward to your replies.
I'm interestet in learning more about using tools/platforms or own programs to find edges in the market. Can anyone recommend how to start, recommend tools and/or simple sample programs/algos that help getting my toe into the cold water slowly and gain some experience ?
I have experiance in C/C++ and use Sierra Chart. It would also be interesting to know if it makses sense to use Sierra Chart ACSIL for that purpose or if it is necessary to write a compülete tool by my own.
Looking forward to your replies.
Before writing your own check out out google's tensor flow. it does it all or for you you don't even have to normalize the data. Think this algo would be a good one and it was apparently a big deal at the time. they found that short date vol is relatively easy to predict and developed a classification algorithm to predict tomorrows vol then bought or sold the VXX ETF (the one that tries to match the return of around 1m VIX future). The statistic was extremely strong like >75% and it made insane returns and then someone wrote about it and it stopped. However the features they used sounded a bit weird and obviously it's an outdate optimisation. Maybe like 2 realised vol time frames, the short term vix, the vix and 3 vix futures, and the time series data of spx for previous month as inputs. Here's the original:
http://www.iosrjournals.org/iosr-jbm/papers/Vol4-issue4/E0443035.pdf
Def think a lot to do in derivatives since we didn't have data for a long time. What do you do with your C/C+?
http://www.iosrjournals.org/iosr-jbm/papers/Vol4-issue4/E0443035.pdf
Def think a lot to do in derivatives since we didn't have data for a long time. What do you do with your C/C+?