@masked_stat 4/ For live monitoring โ what metrics do you track to detect strategy degradation? Rolling WR, PF, max DD vs backtest? Any specific thresholds that trigger a pause?
Really appreciate the transparency. The feature dictionary you shared is exceptionally well structured. Thanks!
@masked_stat 3/ In your full NT8 version, how many of the 91 features from the dictionary are active as filters? Pine simplified seems to cover ~20-30. Is the gap meaningful in live performance, or do most features have marginal contribution after the top ones?
@masked_stat results from mine or if you trained on that Samp, and you did it without filters, but in this way you have to optimize the parameters frequently. This is clearly for the original monkey, for the ultrainstinct I haven't checked yet, because you only shared the code of the indicat