DrawData
About

DrawData is a data observatory, not a tipster.

We turn public state and national lottery draw history into interactive, honest analytics. The point of view is structural: lottery draws are independent and random, so every view here describes the past — none claims to predict the future. Coverage spans Powerball, Mega Millions, and Pick 3 / Pick 4 games across Wisconsin, Pennsylvania, New Jersey, Texas, and North Carolina — over a quarter-million draws in all.

Data last updated: Jul 8, 2026

Data sources

Historical draw data is pulled directly from the relevant state lottery publications and from data.ny.gov Open Data (Mega Millions). Winning numbers are public facts, but we want to be explicit: DrawData is not affiliated with, endorsed by, or sponsored by the Wisconsin, Pennsylvania, New Jersey, Texas, or North Carolina Lotteries, the Multi-State Lottery Association, or any other official lottery.

CSVs live under data/<state>/ in the project and are processed at build time by npm run ingest. Every dataset currently in the system:

  • Wisconsin Pick 3 — 14,541 draws, 1992-09-21 → 2026-07-08.
  • Wisconsin Pick 4 — 12,714 draws, 1997-09-15 → 2026-07-08.
  • Pennsylvania Pick 3 — 25,942 draws, 1977-03-01 → 2026-07-08.
  • Pennsylvania Pick 4 — 23,431 draws, 1980-11-22 → 2026-07-08.
  • New Jersey Pick 3 — 8,263 draws, 2015-03-09 → 2026-07-08.
  • New Jersey Pick 4 — 8,259 draws, 2015-03-09 → 2026-07-08.
  • Texas Pick 3 — 17,813 draws, 1993-10-25 → 2026-07-08.
  • Texas Daily 4 — 11,753 draws, 2007-10-01 → 2026-07-08.
  • North Carolina Pick 3 — 13,834 draws, 2006-10-06 → 2026-07-08.
  • North Carolina Pick 4 — 11,902 draws, 2009-04-17 → 2026-07-08.
  • Florida Pick 3 — 1,839 draws, 2024-01-01 → 2026-07-08.
  • Florida Pick 4 — 1,839 draws, 2024-01-01 → 2026-07-08.
  • Washington Daily Game — 199 draws, 2025-12-21 → 2026-07-07.
  • Georgia Cash 3 — 25,271 draws, 1993-08-10 → 2026-07-08.
  • Georgia Cash 4 — 22,976 draws, 1997-04-06 → 2026-07-08.
  • Michigan Daily 3 — 20,305 draws, 1998-01-01 → 2026-07-08.
  • Michigan Daily 4 — 20,305 draws, 1998-01-01 → 2026-07-08.
  • Powerball — 3,825 draws, 1992-04-22 → 2026-07-06.
  • Mega Millions — 2,517 draws, 2002-05-17 → 2026-07-07.

Mega Millions and Powerball are national games (same numbers everywhere); their start dates here reflect when our chosen feed begins, not when the game launched. Pennsylvania’s Pick 3 has the deepest single-state record in the system — back to 1977. Texas runs four draws daily; we expose the Day and Night flagship times as Midday/Evening to match the other states.

Powerball is multi-state; its drawing matrix has changed seven times since 1992 (see era table below). Mega Millions changed its matrix again in April 2025 (now 5/70 + 1/24).

Powerball eras

Cross-number analytics in DrawData default to the current matrix only — apples-to-apples comparisons require it. Every draw is tagged with its era at ingest:

1992-04-22 → 1997-11-055/45 + 1/45 (1992–1997)
1997-11-05 → 2002-10-095/49 + 1/42 (1997–2002)
2002-10-09 → 2005-08-315/53 + 1/42 (2002–2005)
2005-08-31 → 2009-01-075/55 + 1/42 (2005–2009)
2009-01-07 → 2012-01-155/59 + 1/39 (2009–2012)
2012-01-15 → 2015-10-075/59 + 1/35 (2012–2015)
2015-10-07 → present5/69 + 1/26 (2015–present)

Methodology

  • Frequency. Raw count per number / digit. The dashed line on every frequency chart is the count you’d expect if the draws were perfectly random (total slots ÷ pool size).
  • Gaps. “Current gap” is the number of draws since a value last appeared. The full gap distribution is geometric (memoryless) under a fair process — we overlay that theoretical curve for comparison.
  • Coverage. Running count of unique values ever drawn. Classic coupon-collector behaviour: fast at first, then slowing.
  • Sums & shapes. Distributions of digit sums and of repeat patterns (doubles, triples, quads). These distributions are mathematical features of independent uniform digits — not strategy signals.
  • Formula Lab. Each user-built rule is evaluated against every consecutive (prev → next) transition. We report straight (exact-position) and box (any-order) hit rates with the chance baseline computed as E[|candidate set|] ÷ outcome space. A split test (first half vs second half) flags overfitting.

What we will never do

  • Claim a chart, ranking, or rule improves your odds of winning.
  • Use language like “due”, “hot pick”, “best play”, or “system that beats the lottery.”
  • Sell tickets, take bets, route to sportsbooks, or run lottery affiliate links.
  • Imitate the branding or layout of any official lottery or government website.

Standing disclaimer

For analysis and entertainment only. Lottery draws are random and independent. This app describes past results and does not predict future numbers, improve your chances of winning, or constitute betting advice. Not affiliated with or endorsed by any official lottery. 18+.