Loaded 831 rows from file_01_full_dataset.csv

══════════════════════════════════════════════════════════════
  Per-quadrant directional accuracy by horizon — FULL SAMPLE
══════════════════════════════════════════════════════════════
Quadrant                          1yr         2yr         3yr         5yr         7yr        10yr
Double Discount             56% (n=50)   53% (n=49)   55% (n=49)   49% (n=39)   48% (n=31)   48% (n=23) 
Underestimated Growth       52% (n=40)   59% (n=37)   59% (n=34)   76% (n=29)   60% (n=25)   57% (n=14) 
Value Trap                  43% (n=68)   49% (n=63)   50% (n=58)   47% (n=38)   41% (n=27)   50% (n=18) 
Expensive Hype              54% (n=120)   54% (n=109)   55% (n=95)   57% (n=79)   61% (n=56)   79% (n=28) 
ALL DIRECTIONAL             51% (n=278)   53% (n=258)   54% (n=236)   56% (n=185)   54% (n=139)   60% (n=83) 

══════════════════════════════════════════════════════════════
  Same matrix — HIGH-QUALITY COMPOUNDER SAMPLE (ROIC > 15%)
══════════════════════════════════════════════════════════════
Quadrant                          1yr         2yr         3yr         5yr         7yr        10yr
Double Discount             57% (n=42)   59% (n=41)   61% (n=41)   55% (n=33)   56% (n=25)   55% (n=20) 
Underestimated Growth       56% (n=34)   65% (n=31)   64% (n=28)   80% (n=25)   64% (n=22)   58% (n=12) 
Value Trap                  47% (n=43)   50% (n=40)   53% (n=38)   54% (n=26)   44% (n=18)   50% (n=12) 
Expensive Hype              56% (n=66)   57% (n=58)   51% (n=51)   54% (n=41)   56% (n=27)   78% (n= 9) 
ALL DIRECTIONAL             54% (n=185)   57% (n=170)   56% (n=158)   59% (n=125)   55% (n=92)   58% (n=53) 

══════════════════════════════════════════════════════════════
  Combined directional accuracy — binomial p-value vs 50% null
══════════════════════════════════════════════════════════════
 horizon     n   hits    acc           95% CI   p (one-sided)
    1yr   278    143    51%  [ 46%,  57%]       0.3373 
    2yr   258    138    53%  [ 47%,  59%]       0.1449 
    3yr   236    128    54%  [ 48%,  61%]       0.1080 
    5yr   185    104    56%  [ 49%,  63%]       0.0528 *
    7yr   139     75    54%  [ 45%,  63%]       0.1982 
   10yr    83     50    60%  [ 49%,  71%]       0.0392 **

  HQ sample (ROIC > 15%):
 horizon     n   hits    acc           95% CI   p (one-sided)
    1yr   185    100    54%  [ 46%,  62%]       0.1517 
    2yr   170     97    57%  [ 49%,  65%]       0.0387 **
    3yr   158     89    56%  [ 49%,  64%]       0.0652 *
    5yr   125     74    59%  [ 50%,  67%]       0.0243 **
    7yr    92     51    55%  [ 46%,  65%]       0.1741 
   10yr    53     31    58%  [ 45%,  72%]       0.1358 

Legend: *** p<0.01, ** p<0.05, * p<0.10

Total observations sampled per horizon: {1: 278, 2: 258, 3: 236, 5: 185, 7: 139, 10: 83}
Pending (forward window > today) per horizon: {1: 0, 2: 20, 3: 42, 5: 93, 7: 139, 10: 195}
