Fundamental principle
The "5 years + 2" methodology is the core of the HYBRIDE engine applied to EuroMillions. It combines three analysis windows — the last 5 years, the last 2 years and the entire available history — to balance statistical stability and responsiveness to trends.
5 years : Frequency analysis over the last 5 years of draws
+ 2 : Analysis of the last 2 years to capture recent trends; as these draws also count in the other windows, they carry more weight
Data sources
- Database : Cloud SQL (MySQL) hosted on Google Cloud
- History : over 700 official EuroMillions draws (last 7 years)
- Frequency : Draws on Tuesdays and Fridays
- Update : Synchronisation after each official draw
- Integrity : Systematic verification of source data
Calculated metrics
Frequencies
For each number (1-50) and each star (1-12), calculation of the number of appearances over the analysed period.
Gaps
Number of draws since the last appearance of each number and each star.
Conformity analysis
Multi-criteria evaluation based on:
- Odd/even balance (optimal ratio: 2-3 even)
- Low/high distribution (numbers 1-25 vs 26-50)
- Dispersion (spread between min and max)
- Total sum (reference range: 94 to 161; outside it, the grid is penalised, not excluded)
- Consecutive sequence detection
- Stars evaluated separately (1-12)
Temporal weighting
The HYBRIDE engine combines three analysis windows, grouped into two layers:
Layer 1 — Long-term base
The last 5 years and the entire available history of EuroMillions draws are analysed to establish reference frequencies and gaps. This base ensures statistical stability and identifies deep trends.
Layer 2 — Recent boost
The last 2 years are analysed separately. As the windows overlap, a recent draw counts in all three and therefore carries more weight. This captures potential trend shifts without losing historical perspective.
The name "5 years + 2" refers to the windows of the last 5 and the last 2 years; the engine adds the entire available history.
Generation process
The HYBRIDE engine follows a 5-step pipeline for each generated grid:
- Collection : Extraction of official draws from the database
- Calculation : Frequencies, gaps and temporal weightings for numbers (1-50) and stars (1-12)
- Generation : Probabilistic selection of candidate combinations (5 numbers + 2 stars)
- Scoring : Multi-criteria evaluation of each candidate combination
- Selection : Ranking by conformity score and returning the best grids
Scientific limitations
Warning : EuroMillions is a game of chance. Each draw is statistically independent. No method can predict future results.
The "5 years + 2" methodology offers statistically guided grids, not predictions.
Probability of winning the jackpot: 1 in 139,838,160
Tech stack
- Backend : FastAPI (Python 3.13+)
- Infrastructure : Google Cloud Run (auto-scaling)
- Database : Cloud SQL (MySQL)
- Engine : HYBRIDE
- Region : Google Cloud — European Union