diff --git a/risk/src/var_calculator/parametric.rs b/risk/src/var_calculator/parametric.rs
index 042d624f9..dc0abef45 100644
--- a/risk/src/var_calculator/parametric.rs
+++ b/risk/src/var_calculator/parametric.rs
@@ -146,24 +146,26 @@ impl ParametricVaR {
             .ok_or_else(|| anyhow::anyhow!("Failed to convert VaR amount to Decimal: {var_amount}"))
     }
 
-    /// Get z-score for given confidence level
+    /// Get z-score for given confidence level using Abramowitz-Stegun
+    /// rational approximation for the inverse normal CDF.
+    /// Accurate to ~4.5e-4 across the full range.
     fn get_z_score(confidence_level: f64) -> f64 {
-        // Approximate z-scores for common confidence levels
-        match (confidence_level * 100.0) as u32 {
-            90 => 1.282,
-            95 => 1.645,
-            99 => 2.326,
-            _ => {
-                // Linear interpolation for other values
-                if confidence_level <= 0.90 {
-                    1.282 * confidence_level / 0.90
-                } else if confidence_level <= 0.95 {
-                    1.282 + (1.645 - 1.282) * (confidence_level - 0.90) / 0.05
-                } else {
-                    1.645 + (2.326 - 1.645) * (confidence_level - 0.95) / 0.04
-                }
-            },
+        let p = 1.0 - confidence_level;
+        if p <= 0.0 || p >= 1.0 {
+            return 2.326; // fallback to 99% z-score for invalid inputs
         }
+
+        let t = (-2.0_f64 * p.ln()).sqrt();
+
+        // Coefficients from Abramowitz-Stegun formula 26.2.23
+        let c0 = 2.515517;
+        let c1 = 0.802853;
+        let c2 = 0.010328;
+        let d1 = 1.432788;
+        let d2 = 0.189269;
+        let d3 = 0.001308;
+
+        t - (c0 + c1 * t + c2 * t * t) / (1.0 + d1 * t + d2 * t * t + d3 * t * t * t)
     }
 
     /// Calculate component `VaR` (marginal contribution to `VaR`)
@@ -228,10 +230,48 @@ mod tests {
     }
 
     #[test]
-    fn test_z_score_calculation() {
-        assert!((ParametricVaR::get_z_score(0.90) - 1.282).abs() < 0.01);
-        assert!((ParametricVaR::get_z_score(0.95) - 1.645).abs() < 0.01);
-        assert!((ParametricVaR::get_z_score(0.99) - 2.326).abs() < 0.01);
+    fn test_z_score_accuracy() {
+        assert!(
+            (ParametricVaR::get_z_score(0.90) - 1.282).abs() < 0.01,
+            "90% z-score: got {}, expected ~1.282",
+            ParametricVaR::get_z_score(0.90)
+        );
+        assert!(
+            (ParametricVaR::get_z_score(0.95) - 1.645).abs() < 0.01,
+            "95% z-score: got {}, expected ~1.645",
+            ParametricVaR::get_z_score(0.95)
+        );
+        assert!(
+            (ParametricVaR::get_z_score(0.99) - 2.326).abs() < 0.01,
+            "99% z-score: got {}, expected ~2.326",
+            ParametricVaR::get_z_score(0.99)
+        );
+        assert!(
+            (ParametricVaR::get_z_score(0.999) - 3.090).abs() < 0.05,
+            "99.9% z-score should be ~3.09, not ~2.48; got {}",
+            ParametricVaR::get_z_score(0.999)
+        );
+    }
+
+    #[test]
+    fn test_z_score_edge_cases() {
+        // 50% confidence should give z near 0
+        let z = ParametricVaR::get_z_score(0.5);
+        assert!(
+            z.abs() < 0.1,
+            "50% z-score should be ~0.0, got {}",
+            z
+        );
+        // Very high confidence should give large z
+        let z_high = ParametricVaR::get_z_score(0.9999);
+        assert!(
+            z_high > 3.0,
+            "99.99% z-score should be >3.0, got {}",
+            z_high
+        );
+        // Boundary: p=0 or p=1 should return fallback
+        assert!((ParametricVaR::get_z_score(1.0) - 2.326).abs() < 1e-6);
+        assert!((ParametricVaR::get_z_score(0.0) - 2.326).abs() < 1e-6);
     }
 
     #[test]