Machine learning in trading: theory, models, practice and algo-trading - page 2375
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Here the increments, without sl and tp
TP = SL = 0.005 this is a fixed increment up or down of the current price. When it is reached, the TP or SL will trigger.
TP = SL = 0.005 this is a fixed increment up or down from the current price. Upon achievement of which the TP or SL will trigger
this is a threshold for buy/sell deals in relation to zero
everything inside - do not tradeHere are the increments, no sls, etc.
I did this through clustering, marked up. In general the curve on marked-up data is not very good, but it is more stable on new dataIt's always the other way around for me. Or you're not comparing with the training set, but with something else?
It's always the other way around for me. Or you are not comparing with the training set, but with something else?
If you cluster by increments (2 or 3 clusters) and use them for trading, then the balance curve is not very good, but stable on new data (after training)
and quality of classification is always close to 1
This material is more interesting. I just don't understand, does it only work from the command line? Has anyone looked at it?
I do not really understand where and what works on the command line. At first sight, the basis of the approach lies in the description of the behavior of the price on a sufficiently long time interval as a dynamic system (stochastic diffusion). I think such approaches are not quite adequate due to the essential non-stationarity of the price - I mean for our main task of searching for alpha.
I really like markdowns based on a zigzag approach (but it has problems there too).
this is the threshold for buy/sell transactions relative to zero
From the current price I mean.
In short, it's a stupid clustering of increments into 3 clusters. If we return to the price and color it, there will be directions of deals. If we open trades on them, there will be not a very smooth balance chart
I have an idea to make a noise filter and mark trades after it in my own way...
this is the threshold for buy/sell trades relative to zero
everything inside - do not tradeThat's what I did - I marked 0 with the inactivity class
That's what I did - I marked 0 class of inactivity
Well I had a pretty good akurasi on the new data, but the balance curves weren't great.
I.e. mark the deals by some waveform with a period of 5 or the price difference and see what happens
The signs will also be smoothed during the training