Browse the archive
Filter by game, date, session, number, and rules era.
Evidence-first desktop research
See the record clearly.
Explore historical lottery drawings, inspect their provenance, and test pattern claims against later untouched results—without pretending the past predicts the next draw.
DrawScope does not predict winning numbers, improve lottery odds, or provide betting advice.
A practical research workflow
What you can do
Move from archive inspection to a bounded conclusion without hiding uncertainty.
Inspect provenance
See coverage, source identity, hashes, verification state, and known gaps.
Test a fixed claim
Apply the same ranking rule only to information available before each target.
Protect confirmation
Select on discovery data once, then measure on later untouched trials.
Compare with chance
Use seeded baselines, lift, stability blocks, and significance estimates.
Read the boundary
Keep exact game odds separate from historical evidence and recommendations.
Guided product tour
Three views. One evidence trail.
A real packaged run
The honest result is often “no demonstrated edge.”
This evidence is captured through the packaged DrawScope executable and analytics sidecar, using the bundled archive and strict contract boundary.
Bounded historical confidence
5/ 100
Retrospective-only cap: 49/100
No demonstrated predictive advantage
- Confirmation lift
- 1.16×
- One-sided p-value
- 0.153
- Tested trials
- 250
- Recommendation
- do not use to choose numbers
Trace the result, not just the screenshot.
The checked-in evidence bundle records the application version, database hash, methodology, request boundary, full structured result, and reproduction steps.
- The target draw never participates in the ranking used to evaluate it.
- The strongest discovery strategy is measured once on later confirmation data.
- The confidence score remains below 50 until separately specified prospective evidence exists.
- Exact lottery odds remain unchanged by the historical result.
Archive status
41,598 drawings with limits attached.
6 games5 sources4 documented gaps
| Game | First draw | Latest captured | Draws | Sessions |
|---|---|---|---|---|
| Powerball | 1992-04-22 | 2026-07-27 | 3,813 | 1 |
| Mega Millions | 2002-05-17 | 2026-07-24 | 2,522 | 1 |
| Illinois Lotto | 2014-01-20 | 2026-07-27 | 1,960 | 1 |
| Lucky Day Lotto | 2014-01-19 | 2026-07-28 | 9,147 | 2 |
| Pick 3 | 2010-01-01 | 2026-07-28 | 12,078 | 2 |
| Pick 4 | 2010-01-01 | 2026-07-28 | 12,078 | 2 |
Database identity
89a9370d4dcbba7a6ca22e218e4ed6ba6ff1a960b5c1247f3f3f4a0a4569662f
Keep the limits visible
Patterns are descriptions, not promises.
Independent lottery drawings retain their theoretical odds. DrawScope helps test whether an apparent historical relationship survives a forward-only evaluation; it does not turn that relationship into a forecast.
Local-firstNo account, telemetry, or cloud analytics
ReproducibleHash-checked sources and byte-identical database rebuilds
AuditableStrict React ↔ Rust ↔ Python contracts
VerifiableChecksums, SBOM, provenance, CI, and CodeQL
Windows x64 · v0.6.5
Explore the record on your own machine.
The current release is a portable ZIP. Verify its checksum, extract it to a writable folder, and run launch-portable.bat.
The current public binary is not Authenticode-signed. The repository now requires signing credentials before future installer releases can publish.