How Scratchstats ranks scratch-off games.
A short orientation followed by reference material. Covers the EV math, the state feeds we ingest, the grading rubric, and where our data falls short.
Overview.
A short orientation. Read this first; the rest is reference.
ScratchStats pulls prize data from thirteen state lotteries, scores every active game against the same rubric, and ranks them on the Big Board so you can see at a glance which tickets are still paying and which have already been picked clean.
EV — worked example.
Pick a prize tier; the math updates with that tier's actual numbers.
Expected value is the sum of every prize tier's contribution to a single ticket's payout, divided by ticket price. Tap a tier in the table below and the formula on the right will update with that tier's exact numbers.
Coverage by state.
Thirteen states. Different cadences, different feed shapes.
Grading rubric.
One letter per game. A is rare on purpose.
Material structural improvement vs. launch. The handful of games we'd actually point at.
Tracking near or slightly above launch. The default state for most games most of the time.
Materially worse than at launch. Either skip or wait for a reset.
Limitations.
The honest caveats. Read these before you act on the numbers.
- 01
Prize-claim data lags real-world purchases by the state's publish cadence. A game can be hot for three days before a weekly feed reflects it.
- 02
Estimated tickets remaining is exactly that, an estimate disclosed per game when used. The closer a game is to retirement, the wider the error band.
- 03
EV is not a perfect indicator and is heavily skewed by outliers. It does not mean a player will win X% of the time, or get back X% of every dollar they spend; one $10M top prize sitting unclaimed can lift a game's EV more than thousands of small prizes ever will.
- 04
Winning tickets get pocketed, lost, and forgotten. State lotteries report tens of millions of dollars in unclaimed prizes every year (Massachusetts alone regularly tops $30M annually). Our 'tickets remaining' counts assume those winners eventually cash; in practice a small but real share never do.