Tweak the backtests.

* Start from 2011 instead of 2013.
* Validate over two years instead of one.

NOTES.md

@@ -112,7 +112,7 @@ After running comprehensive backtests across multiple periods from 2013-2024, we
 
 1. Refined Backtest Framework
 - Implemented more granular 6-month step testing periods
-- Standardized minimum training period (1 years) and validation window (8 years)
+- Standardized minimum training period (2 years) and validation window (8 years)
 - Separated results into "normal" and "stress" periods for clearer performance assessment
 
 2. Performance Evaluation Approach

model.py

@@ -1019,6 +1019,8 @@ def run_projection(args):
 def run_projections(df):
     # Create main projection
     projection_starts = [
+        "2011-01-01",
+        "2012-01-01",
         "2013-01-01",
         "2014-01-01",
         "2015-01-01",
@@ -1073,7 +1075,7 @@ def run_single_backtest(args):
         return {"params": params, "error": str(e), "success": False}
 
 
-def run_systematic_backtests(df, validation_years=1, min_training_years=8):
+def run_systematic_backtests(df, validation_years=2, min_training_years=8):
     """
     Run a comprehensive suite of backtests with consistent validation periods.
     Uses sliding windows for both start and end dates.
@@ -1083,7 +1085,7 @@ def run_systematic_backtests(df, validation_years=1, min_training_years=8):
     min_training_days = min_training_years * 365
 
     # Define start date for reliable data
-    mature_start = pd.Timestamp("2013-01-01")
+    mature_start = pd.Timestamp("2011-01-01")
     last_possible_start = df["Date"].max() - pd.Timedelta(
         days=min_training_days + validation_days
     )