Records (VALID rows with 5y forward returns + all primitives): 216

==========================================================================================
 HEAD-TO-HEAD: 8 value signals on identical data, 2014-2024
 Tercile method: top-1/3 cheap → predict outperform, bottom-1/3 expensive → underperform
==========================================================================================
  P/E             n= 138  hits= 65  acc=47.1%  CI [ 38%,  55%]  p=0.7781 
  EV/EBITDA       n= 138  hits= 69  acc=50.0%  CI [ 42%,  58%]  p=0.5339 
  P/B             n= 128  hits= 58  acc=45.3%  CI [ 37%,  54%]  p=0.8748 
  FCF yield       n= 116  hits= 63  acc=54.3%  CI [ 45%,  64%]  p=0.2017 
  Earnings yld    n= 138  hits= 65  acc=47.1%  CI [ 38%,  55%]  p=0.7781 
  EV/NOPAT        n= 138  hits= 70  acc=50.7%  CI [ 42%,  59%]  p=0.4661 
  OE-DCF MoS      n= 138  hits= 74  acc=53.6%  CI [ 46%,  62%]  p=0.2219 
  Brina Gap       n= 138  hits= 80  acc=58.0%  CI [ 50%,  66%]  p=0.0367 **

==========================================================================================
 SAME COMPARISON, split by long-side / short-side
==========================================================================================
  P/E             total n=138  acc=47.1% [ 39%,  55%] p=0.7781      LONG (cheap) n= 69 acc= 46%     SHORT (expensive) n= 69 acc= 48%
  EV/EBITDA       total n=138  acc=50.0% [ 41%,  59%] p=0.5339      LONG (cheap) n= 69 acc= 48%     SHORT (expensive) n= 69 acc= 52%
  P/B             total n=128  acc=45.3% [ 37%,  54%] p=0.8748      LONG (cheap) n= 64 acc= 48%     SHORT (expensive) n= 64 acc= 42%
  FCF yield       total n=116  acc=54.3% [ 46%,  64%] p=0.2017      LONG (cheap) n= 58 acc= 52%     SHORT (expensive) n= 58 acc= 57%
  Earnings yld    total n=138  acc=47.1% [ 39%,  55%] p=0.7781      LONG (cheap) n= 69 acc= 46%     SHORT (expensive) n= 69 acc= 48%
  EV/NOPAT        total n=138  acc=50.7% [ 43%,  59%] p=0.4661      LONG (cheap) n= 69 acc= 51%     SHORT (expensive) n= 69 acc= 51%
  OE-DCF MoS      total n=138  acc=53.6% [ 46%,  62%] p=0.2219      LONG (cheap) n= 69 acc= 54%     SHORT (expensive) n= 69 acc= 54%
  Brina Gap       total n=138  acc=58.0% [ 49%,  66%] p=0.0367 **     LONG (cheap) n= 69 acc= 57%     SHORT (expensive) n= 69 acc= 59%

==========================================================================================
 INTERPRETATION GUIDE
==========================================================================================
 - Tercile-based signal test is the conventional factor-portfolio framework.
 - All signals tested on the same set of underlying observations.
 - 'Higher accuracy' = the signal better identifies outperformers (long side)
   AND/OR better identifies underperformers (short side).
 - Practitioner takeaway: if Brina Gap is on top, you have a compelling
   improvement over the conventional value signals readers already trust.
