Public data with traceable sources.
Both the ground truth and the signals come from public data whose source can be named.
Sources
Data and use
| Data | Use | Frequency |
|---|---|---|
| aT Yangjae flower auction (OpenAPI) | Korean cut-flower prices (ground truth) | Daily |
| KAMIS · aT Seoul Garak wholesale | Produce prices (ground truth) | Daily |
| UN Comtrade | Import value ÷ net weight by origin (ground truth) | Monthly |
| World Bank Pink Sheet | Commodity benchmark prices (ground truth) | Monthly |
| ESA Sentinel-2 L2A · WorldCover | 10 m crop supply index in main regions (cropland only) | 5 days |
| NASA MODIS MOD13Q1 | 250 m vegetation index (NDVI) | 16 days |
| ERA5 · Open-Meteo | Temperature, rainfall and sunlight anomalies, 16-day forecast | Daily |
| ECB (frankfurter) | KRW, USD, EUR, JPY conversion | Daily |
Sources as labelled in the BLOOMCAST console and product page. Frequency is that of the original data.
Principle
Ground truth and signals are kept apart
Ground truth
Only real traded prices (auction, wholesale, customs, benchmarks) are used as targets.
Signals
Satellite, weather, exchange rates and the demand calendar explain prices and are used only when the backtest shows they help.
Timing
Backtests use only the data actually available at each date.
FAQ
Frequently asked questions
Can we add our own data?
On Enterprise we build a dedicated model using your purchase or auction records as ground truth.
Related pages
Get started
Tell us the items you trade.
We reply with a trial forecast for those items, shown next to the simple “price stays the same” baseline, so you can judge it on your own numbers.