fix(critical): complete Zt sign alignment across all modules

Fixed ALL instances of (d - sqrt_rho*z) -> (d + sqrt_rho*z):
- models/vasicek.py: conditional_transition_matrix() (used by lifetime PD)
- data/transition_matrices.py: _generate_model_consistent_matrix()
- models/credit_cycle.py: already fixed in previous commit

Added sign convention docs:
- vasicek.py conditional_pd() uses Basel convention (Z↑=loss↑)
- conditional_transition_matrix() uses Belkin convention (Z↑=호황)
- Both conventions documented in module docstrings

Pipeline 8/8 validation pass after fix

data/transition_matrices.py

@@ -128,18 +128,18 @@ def _generate_model_consistent_matrix(
             cum_prob_clipped = np.clip(cum_prob, 1e-10, 1.0 - 1e-10)
             thresholds[i, j] = norm.ppf(cum_prob_clipped)
     
-    # 2. Z-조건부 전이확률 계산
+    # 2. Z-조건부 전이확률 계산 (Belkin convention: Z>0 = 호황)
     cond_tm = np.zeros((n, n))
     for i in range(n - 1):
         for j in range(n):
             d_upper = thresholds[i, j]
-            upper = norm.cdf((d_upper - sqrt_rho * z) / sqrt_1_rho)
+            upper = norm.cdf((d_upper + sqrt_rho * z) / sqrt_1_rho)
             
             if j == 0:
                 lower = 0.0
             else:
                 d_lower = thresholds[i, j - 1]
-                lower = norm.cdf((d_lower - sqrt_rho * z) / sqrt_1_rho)
+                lower = norm.cdf((d_lower + sqrt_rho * z) / sqrt_1_rho)
             
             cond_tm[i, j] = max(upper - lower, 0.0)
         

models/vasicek.py

@@ -1,9 +1,12 @@
 """
 Vasicek 단일팩터 모델 기반 조건부 PD 및 전이행렬 모듈
 
-핵심 공식:
+핵심 공식 (Basel/Vasicek convention: Z↑ = loss↑ = 불황):
   PD_PIT(Z) = Φ( (Φ⁻¹(PD_TTC) - √ρ · Z) / √(1-ρ) )
 
+주의: Belkin & Suchower에서는 Z↑ = 호황 (반대 부호).
+  조건부 전이행렬은 Belkin convention 사용 (d + √ρ·Z).
+
 이 모듈은 Belkin & Suchower의 임계값 방식 대신, 
 Vasicek 공식을 직접 적용하는 간편 버전도 제공합니다.
 
@@ -27,6 +30,9 @@ def conditional_pd(pd_ttc: float, z: float, rho: float) -> float:
     
     PD_PIT(Z) = Φ( (Φ⁻¹(PD_TTC) - √ρ · Z) / √(1-ρ) )
     
+    주의: 이 함수의 Z는 Basel/Vasicek convention (Z↑ = 불황).
+    Belkin Z(양수=호황)를 사용하려면 -Z를 넣어야 합니다.
+    
     Parameters
     ----------
     pd_ttc : float  - TTC (Through-the-Cycle) 부도확률
@@ -110,18 +116,19 @@ def conditional_transition_matrix(
             thresholds[i, j] = norm.ppf(cum_prob_clipped)
     
     # 조건부 전이행렬 계산
+    # Belkin convention: Z>0 = 호황, 누적확률 오름차순 → (d + √ρ·Z)
     cond_tm = np.zeros((n, n))
     
     for i in range(n - 1):
         for j in range(n):
             d_upper = thresholds[i, j]
-            upper = norm.cdf((d_upper - sqrt_rho * z) / sqrt_1_rho)
+            upper = norm.cdf((d_upper + sqrt_rho * z) / sqrt_1_rho)
             
             if j == 0:
                 lower = 0.0
             else:
                 d_lower = thresholds[i, j - 1]
-                lower = norm.cdf((d_lower - sqrt_rho * z) / sqrt_1_rho)
+                lower = norm.cdf((d_lower + sqrt_rho * z) / sqrt_1_rho)
             
             cond_tm[i, j] = max(upper - lower, 0.0)