feat: forecast start date defaults to next open market date

Signed-off-by: ivelin <[email protected]>

Author: ivelin

forecast.sh

@@ -1,9 +1,12 @@
 #!/usr/bin/bash
+# Use this script to run forecast for multiple periods
+# with custom forecast start dates
 echo "Running forecast for multiple periods"
 set -exv
 
-
+# without setting a specific forecast start date, forecast will start from the next open market date
 #./canswim.sh forecast
+
 #./canswim.sh forecast --forecast_start_date "2023-11-20"
 #./canswim.sh forecast --forecast_start_date "2023-12-04"
 #./canswim.sh forecast --forecast_start_date "2023-12-18"
@@ -50,4 +53,5 @@ set -exv
 #./canswim.sh forecast --forecast_start_date "2024-11-25"
 #./canswim.sh forecast --forecast_start_date "2024-12-02"
 #./canswim.sh forecast --forecast_start_date "2024-12-09"
-./canswim.sh forecast --forecast_start_date "2024-12-16"
+#./canswim.sh forecast --forecast_start_date "2024-12-16"
+#./canswim.sh forecast --forecast_start_date "2024-12-23"

setup.cfg

@@ -24,6 +24,7 @@ python_requires = >=3.10
 install_requires =
     darts==0.27.2
     yfinance[nospam]
+    pandas_market_calendars
     fmpsdk
     optuna==3.4.0
     huggingface_hub

src/canswim/forecast.py

@@ -12,6 +12,8 @@
 from canswim import constants
 from typing import List
 import duckdb
+from datetime import datetime, timedelta
+import pandas_market_calendars as mcal
 
 
 class CanswimForecaster:
@@ -233,12 +235,44 @@ def upload_data(self):
         self.hfhub.upload_data()
 
 
+
+def get_next_open_market_day():
+    # Get calendar for NYSE
+    nyse = mcal.get_calendar('NYSE')
+    
+    # Get today's date
+    today = datetime.now().date()
+    
+    # Look for the next valid trading day within a reasonably big window of 30 days
+    valid_days = nyse.valid_days(start_date=today, end_date=today + timedelta(days=30))
+    
+    next_trading_day = None
+
+    if valid_days is not None and len(valid_days) > 0:
+        next_trading_day = valid_days[0]
+
+    if next_trading_day is not None:
+        logger.info(f"The next open stock market date is: {next_trading_day}")
+    else:
+        logger.warning("No open market day found within the next 30 days.")
+
+    # If we can't find a next valid day within 30 days (which shouldn't happen for NYSE), return None
+    return next_trading_day
+
+
 # main function
 def main(forecast_start_date: str = None):
     logger.info("Running forecast on stocks and uploading results to HF Hub...")
     if forecast_start_date is not None:
         logger.info(f"forecast_start_date: {forecast_start_date}")
         forecast_start_date = pd.Timestamp(forecast_start_date)
+    else:
+        # get next open stock market date
+        # Example usage
+        forecast_start_date = get_next_open_market_day()
+
+    logger.info(f"Forecast start date set to: {forecast_start_date}")
+
     cf = CanswimForecaster()
     cf.download_model()
     cf.download_data()