Settings & Data Management
Upload historical data and train prediction models
Upload Historical Data
Upload a CSV file with historical volume data. Supports two formats:
Format 1: Daily Aggregates
Date,Volume
2025-01-15,28450
2025-01-16,31200
Format 2: Intraday (15-min bins)
Date,Time,Volume,Open,High,Low,Close
2025-01-15,10:45,1250,245.50,246.00,245.25,245.75
2025-01-15,11:00,2100,245.75,246.25,245.50,246.00
Model Training
Train or retrain both prediction models (HCVP and CMEM) using the historical data in the database. Training is automatically triggered after CSV upload, but you can manually retrain if needed.
Database Information
Database Type
SQLite
Location
packages/database/dev.db
Training Window
90 days
Confidence Level
95%
API Endpoint: http://localhost:3001
Dashboard: http://localhost:3000
Live Data Scraper
Scraper Configured & Active
The scraper is currently configured to fetch volume data directly from Euronext. No manual configuration is required.
Strategy
PlaywrightHeadless
Target Table
#future-prices-tableRow SelectionNearest Contract
Logic: First row in table body (tr:first-child)
Column SelectionDay Volume
Logic: 7th column (td:nth-child(7))
Update FrequencyReal-time (On Prediction Request)
Last Value Read
N/A
Calendar Effects
Adjust volume multipliers for special trading days. These affect the CMEM model's static forecast. A multiplier of 1.0 = no adjustment, 1.2 = +20% volume, 0.7 = -30% volume.
USDA Report
WASDE / Grain Stocks release days (+volume at 18:00 CET)
+20%
Roll Period
5 days before contract expiry (Mar, May, Sep, Dec)
+50%
Harvest Season
July-August European wheat harvest
+15%
US Holiday
NYSE/CME closed, reduced cross-market activity
-30%
French Holiday
Euronext Paris closed
-100%
Standard
Normal trading day (no adjustments)
+0%
Quick Start
- 1.Upload your historical MATIF Wheat volume data (CSV format)
- 2.Wait for automatic training to complete (or click "Train Models")
- 3.Go to the Dashboard to see predictions
- 4.Check the Backtest page to evaluate model accuracy