Raw and source-shaped rows across lottery, macro, calendar, geography, and weather.
Prediction intelligence workbench
Data Lake Command Center
A sharp analytical system for cutting through raw data, building lakehouse features, and reporting prediction uncertainty with discipline.
Validated Mega-Sena features with cadence, calendar, distribution, and draw metadata.
Model-ready rows joined to weather, holidays, and macroeconomic context.
DuckDB views mapped onto Parquet tables for instant local analytical SQL.
Lineage
From raw source to model input
- Caixa3017 contests
- Bronzemegasena_draws.parquet
- Silverfeature synthesis
- Goldmodel_input.parquet
- DuckDBgold_megasena_model_input
Baseline backtest
Frequency plus recency
Average best hits: 1.5244 over 2,967 evaluated draws. This is an honesty anchor, not a promise of predictability.
Collectors
Context lake
Candidate tickets
Current baseline output
Lottery outcomes are random by design. These candidates are analysis output, not betting advice.