Subject: Re: NHNL Calculation
As a quick double-check of the OP results, I changed one line in the standard BCC code for [NHNLDiff] to use nas100.a=1. The results were similar to the OP: improvement in all metrics when applied to N1T. Results from 19851231 to 20251128:

  Screen     CAGR  SAWR  GSD  MDD  Sharpe  AT
    N1T      14.7  4.7   26   -81   0.58   0.0
  N1T_BCC0   15.7  6.0   23   -72   0.66   0.6
N1T_BCCN10   16.1  8.0   23   -60   0.68   1.0
 
    S5T      11.5  5.7   18   -55   0.58   0.0
  S5T_BCC0   12.8  7.9   15   -34   0.73   0.6
S5T_BCCN10   12.7  8.3   15   -34   0.72   1.0
 
    S5TE     11.9  8.2   20   -60   0.55   0.0
 S5TE_BCC0   13.3  10.4  17   -39   0.70   0.6
S5TE_BCCN10  13.1  10.4  17   -39   0.68   1.0

BCCN10 timing uses nas100 NHNL.
BCC0 timing uses nasdaq NHNL.

BCCN10 applied to S5T or S5TE lowers CAGR slightly (compared to BCC0), with similar risk metrics. There is more turnover with BCCN10.

I did not try to fine tune the NHNL cutoff. The standard BCC NHNLDiff cutoff of -0.4 corresponds to about -18 in the OP table (multiply by 45, the sum of the weights).

https://gtr1.net/2013/?~N1T:h1...
https://gtr1.net/2013/?~N1T_BC...
https://gtr1.net/2013/?~N1T_BC...