This is a modification of my Scaled Normalized Vector Strategy. This mod features some activation functions. Performance remains high. The repainting problem should be tested out.
To avoid repainting it is recommended to use these lines in the getdiff function (marked with *** on both sides):
curr = scaleMinimax((useold ? security(syminfo.tickerid, tostring(tf), ***hlc3[1]***, barmerge.gaps_off, barmerge.lookahead_on)
: security(syminfo.tickerid, tostring(tf), ***hlc3[1]***)), tf, 0, 1)
Mind the commenting subsystem deletes square brackets, so literally the bit ***hlc3[1]*** is hlc3 + square brackets + 1 + square brackets.
TY19
⋅
@capissimo, Repaint still happened in my 1M chart around 4pm though I changed it to hlc3 as mentioned.
Thank you for this brilliant strategy. I think the repaint problem comes from the chart itself! There are many missing FRAMES in each chart, specially shorter TFs. Karobein is a pattern predictor so it must change the pattern when some frames are added later to the chart.
Goldmineindicator
⋅
Hello Bro,
Really appreciate your work but its repaints lots...
MarcoCons
⋅
What are the best parameters for the 15 min timeframe?
dc4free
⋅
Been looking through your work and seems like a lot of your strategies repaint, man.
curr = scaleMinimax((useold ? security(syminfo.tickerid, tostring(tf), ***hlc3[1]***, barmerge.gaps_off, barmerge.lookahead_on)
: security(syminfo.tickerid, tostring(tf), ***hlc3[1]***)), tf, 0, 1)
Mind the commenting subsystem deletes square brackets, so literally the bit ***hlc3[1]*** is hlc3 + square brackets + 1 + square brackets.