In short: Auto Pick is an Explorer (Study) tool. It does not invent new numbers — it gathers already computed sets from other sections and ranks the best three by your criterion order and weights. It does not change ticket odds and does not predict tomorrow's draw.
1. Why this section exists
Verdario has many independent loops: Pools, Scenarios, Numbers, intersections/unions. Comparing dozens of precalc variants by hand is slow.
Auto Pick does one job:
collect candidates → normalize metrics → weight them → show the top 3 with a live backtest and an “Add to basket” button.
2. Access
| Who | What they see |
|---|---|
| Guest / tiers 1–2 | The menu item is visible; inside — a Study lock and a short note |
| Explorer (Study) | Full tool: size range, depth, criteria, weights, top 3 |
Without Study the section is a showcase: it shows the tool exists but does not run scoring.
3. Where candidates come from
Auto Pick reads precalc (and session combine results when present) at the chosen backtest depth:
- Pools — curated sets by archive year.
- Scenarios (signals) — Stable/Chaos phase-model sets.
- Board · Standard — numbers-board recommendation methods.
- Board · Advanced / Rec OR — advanced board modes.
- Intersections (intersect) — scenario-family agreements (∩).
- Intersect5 — variants aimed at exact 5-hits.
- Unions (union) — wider combinations of loops (∪).
Only sets sized 18–48 that fall inside your from–to range are kept.
If there are no candidates — usually precalc is missing for that depth/size, or the range is too narrow.
4. The three scoring metrics
Every candidate is compared on the same three axes:
| Metric | Meaning |
|---|---|
| Prediction efficiency | Lift×100% on 6-hits vs a random set of the same size. 100% ≈ chance. |
| Rating | Integrated backtest (weights for 6 / 5 / 4 / 3 hits). |
| Cost efficiency | How “expensive” the set size is for the hit quality at that depth. |
Before weighting, each metric is min–max normalized across the current candidate pool. Then your criterion order and weights are applied.
5. All section features
A. Set size from / to
Range 18–48. Only candidates inside this window are kept.
B. Backtest depth
10 / 20 / 30 (by subscription; Study → up to 30).
Depth chooses which precalc slice is read and which live backtest is drawn under the card.
C. Criterion order
Three presets:
- Pred. efficiency → Rating → Cost efficiency
- Cost efficiency → Pred. efficiency → Rating
- Rating → Pred. efficiency → Cost efficiency
Order = axis priority when scoring.
D. Weights for 1st / 2nd / 3rd criterion
Default 1.00 / 0.50 / 0.25.
Weights are normalized at score time (they need not sum to 1). Zero on an axis disables its contribution.
E. “Pick best” button
Collects candidates, computes scores, shows up to three unique sets (identical number sets are deduplicated).
F. Result card
- source label and archive year;
- set balls;
- pred. efficiency / rating / cost efficiency / Score;
- walk-back backtest table: for each draw the prediction is rebuilt from data available at that step (pool / scenario / board / ∩∪), not the final set copied onto every day;
- “Add to basket”.
6. Typical Study workflow
- Open Auto Pick.
- Set a size range, e.g. 24–40.
- Depth 20 or 30 (more stable than 10 for ranking).
- Choose criterion order for your goal:
- beat chance → prioritize pred. efficiency;
- watch set-size cost → prioritize cost efficiency;
- want a stable integrated signal → prioritize rating.
- Keep default weights or boost the first axis (e.g. 1.0 / 0.3 / 0.2).
- Pick best → compare the top 3 → send the best set to the basket.
- In the Basket build a ticket / system — remembering: a 6-number combination keeps the same odds.
7. How it relates to other sections
| Section | Role next to Auto Pick |
|---|---|
| Pools / Scenarios / Numbers | Sources of precalc candidates |
| Method Comparison | Manual AND/OR consensus; Auto Pick auto-ranks finished sets |
| Basket | Where chosen top sets go |
Auto Pick does not replace reading the source sections: it speeds selection when precalc is already computed.
Important to remember
- Auto Pick requires Explorer (Study).
- It ranks already computed sets — it is not a secret new method.
- Score depends on your criterion order and weights — another preset yields another top 3.
- Do not look at Score alone: check pred. efficiency, rating and the backtest table.
- Depth ≥20 is more stable than 10.
- No auto-pick makes a 6-number combination more likely.