diff --git a/src/quant_engine/snapshot_admin.db b/src/quant_engine/snapshot_admin.db
index 25ecc8b..d618e2b 100644
Binary files a/src/quant_engine/snapshot_admin.db and b/src/quant_engine/snapshot_admin.db differ
diff --git a/tools/fix_settings_schema.py b/tools/fix_settings_schema.py
new file mode 100644
index 0000000..2e47078
--- /dev/null
+++ b/tools/fix_settings_schema.py
@@ -0,0 +1,89 @@
+#!/usr/bin/env python3
+"""
+settings 테이블 스키마 수정
+올바른 스키마: ordinal, key, value_json, note, updated_at
+"""
+
+import sqlite3
+import json
+from pathlib import Path
+from datetime import datetime
+
+def fix_settings_schema():
+    """settings 테이블 스키마 수정"""
+
+    db_path = Path('src/quant_engine/snapshot_admin.db')
+    conn = sqlite3.connect(db_path)
+    cursor = conn.cursor()
+
+    # 현재 settings 데이터 백업
+    cursor.execute("SELECT * FROM settings")
+    current_rows = cursor.fetchall()
+
+    print(f"현재 settings: {len(current_rows)}개 행")
+
+    # 현재 컬럼 확인
+    cursor.execute("PRAGMA table_info(settings)")
+    current_cols = cursor.fetchall()
+    print(f"현재 컬럼: {[col[1] for col in current_cols]}")
+
+    # 기존 테이블 삭제
+    cursor.execute("DROP TABLE IF EXISTS settings")
+
+    # 올바른 스키마로 생성
+    cursor.execute("""
+        CREATE TABLE settings (
+            ordinal INTEGER PRIMARY KEY,
+            key TEXT NOT NULL,
+            value_json TEXT NOT NULL,
+            note TEXT DEFAULT '',
+            updated_at TEXT NOT NULL
+        )
+    """)
+
+    print("\n새 스키마 생성:")
+    print("  ordinal INTEGER PRIMARY KEY")
+    print("  key TEXT NOT NULL")
+    print("  value_json TEXT NOT NULL")
+    print("  note TEXT DEFAULT ''")
+    print("  updated_at TEXT NOT NULL")
+
+    # 데이터 복원
+    # 현재 컬럼: [key, value]
+    timestamp = datetime.now().isoformat()
+
+    inserted = 0
+    for ordinal, (key, value) in enumerate(current_rows, start=1):
+        # value를 JSON으로 변환
+        try:
+            value_json = json.dumps(str(value), ensure_ascii=False)
+        except:
+            value_json = json.dumps("", ensure_ascii=False)
+
+        cursor.execute(
+            """INSERT INTO settings (ordinal, key, value_json, note, updated_at)
+               VALUES (?, ?, ?, ?, ?)""",
+            (ordinal, key, value_json, "", timestamp)
+        )
+        inserted += 1
+
+    conn.commit()
+
+    # 검증
+    cursor.execute("SELECT COUNT(*) FROM settings")
+    count = cursor.fetchone()[0]
+    print(f"\n복원된 settings: {count}개 행")
+
+    # 샘플 확인
+    cursor.execute("SELECT ordinal, key, value_json FROM settings LIMIT 3")
+    print("\n샘플 데이터:")
+    for ordinal, key, value_json in cursor.fetchall():
+        value = json.loads(value_json)
+        print(f"  {ordinal}. {key} = {value}")
+
+    conn.close()
+
+    print("\n[OK] settings 테이블 스키마 수정 완료")
+
+if __name__ == "__main__":
+    fix_settings_schema()
diff --git a/tools/load_from_xlsx_correct.py b/tools/load_from_xlsx_correct.py
index 3eae912..7f1d87c 100644
--- a/tools/load_from_xlsx_correct.py
+++ b/tools/load_from_xlsx_correct.py
@@ -49,7 +49,13 @@ class CorrectXLSXLoader:
             header_param = header_row_1based - 1  # pandas는 0-indexed
 
             try:
-                df = pd.read_excel(self.xlsx_file, sheet_name=sheet_name, header=header_param)
+                # settings 특수 처리: 헤더가 없음 (key-value 쌍)
+                if sheet_name == 'settings':
+                    df = pd.read_excel(self.xlsx_file, sheet_name=sheet_name, header=None)
+                    df.columns = ['key', 'value', 'note1', 'note2']
+                    df = df[['key', 'value']]  # 필요한 컬럼만
+                else:
+                    df = pd.read_excel(self.xlsx_file, sheet_name=sheet_name, header=header_param)
 
                 # NaN을 None으로 변환
                 df = df.where(pd.notna(df), None)