Methodology

The EV/Dollar formula

Expected Value per Dollar (EV/Dollar) is the statistical average return per $1 spent on a given scratch-off game, based on the prizes still remaining and the tickets still in circulation. It is computed identically for every game in every state we cover.

tickets_remaining = remaining_winners × overall_odds

EV/Dollar = sum(prize × remaining) / tickets_remaining / ticket_price

• remaining_winners — total prize-winning tickets still unclaimed, summed across all prize tiers, as published by the state lottery.
• overall_odds — the published "1 in N" chance of any prize on the ticket. Multiplied by remaining_winners to back-derive the total tickets still in circulation.
• prize × remaining — for each prize tier, the dollar value of the prize multiplied by the count of tickets still claiming it. Summed across all tiers gives the total prize dollars left on the shelf.
• ticket_price — the printed price of one ticket ($1, $2, $3, $5, $10, $20, $25, $30, or $50).

An EV/Dollar of $1.23 means that, statistically, for every $1 spent on this game today, the average return is $1.23. This is a long-run average across many tickets — not a per-ticket prediction. Every individual ticket is still a game of chance.

The Prize Ratio

Prize Ratio is a secondary metric that tracks how the distribution of remaining prizes compares to the original print run. It surfaces games where the high-value tiers have been disproportionately depleted vs. the lower tiers.

Prize Ratio = (top_prizes_remaining / top_prizes_original)
            ÷ (total_tickets_remaining / total_tickets_original)

A Prize Ratio > 1.0 means the top tier is depleting slower than tickets overall — proportionally more top prizes still available per ticket sold. A ratio < 1.0 means the opposite. EV/Dollar already accounts for this implicitly; Prize Ratio exposes it as a separate visible signal.

Where the data comes from

The "Source" column links to each state's official lottery website — the same URL each scraper points at for daily data collection.

Claim lag

State lottery commissions update their published remaining-prize counts on different cadences. Some publish in near-real-time after each prize claim; others batch updates daily, weekly, or on-claim-event. Even the fastest publishers operate downstream of physical retail sales, so the data we read is always at least slightly behind real ticket-buying activity.

EV/Dollar is therefore a snapshot of yesterday's state of the game, not a live count. We update once per day, and we print the "Last updated" timestamp on the homepage, FAQ, and dashboard so users can see exactly when the rankings were last refreshed.

Per-state synthesis details

For the eight states classified as Synthetic, here is exactly what is modeled vs. what is published directly:

Limits of this approach

• EV/Dollar is a statistical average across many plays. It does not predict the outcome of any individual ticket.
• The lottery retains a mathematical edge on every scratch-off game. EV/Dollar > $1.00 is rare and indicates a game where high-value prizes are disproportionately under-claimed relative to remaining tickets — not a guarantee of profit.
• Synthetic-tier modeling introduces estimation error. We prefer correct static values plus honest NULLs over fabricated variation; we don't invent precision the lottery doesn't publish.
• All data is downstream of what state lottery commissions publish. If a lottery's feed is delayed, incorrect, or taken offline, our rankings reflect that delay until the feed recovers.
• Static-EV states (MN, CO, MT) cannot show day-over-day EV changes. Their rankings are useful as a baseline comparison but should not be expected to track depletion the way dynamic-state rankings do.

Version history

Current methodology version: v1.1 · Last reviewed: May 6, 2026

1. May 6, 2026v1.1

   Added Pennsylvania and Maine to the Synthetic data-quality classification. Pennsylvania synthesizes lower-tier remaining counts from the observed top-6 claim rate; Maine synthesizes from the lottery's published percent-unsold figure. Both follow the same shape as Iowa, Vermont, Wisconsin, and the other synthesis states already documented on this page.

2. April 29, 2026v1.0

   Initial public methodology page. Documents the canonical EV/Dollar formula, Prize Ratio formula, per-state data-quality classification, claim-lag behavior, and the synthesis approach for the six states whose lotteries don't publish complete tier-level remaining counts.

3. April 19, 2026v0.4

   Vermont scraper rebuilt with a single-synthesized-lower-tier model derived from Total Unclaimed dollars + % Sold + total tickets. Added an EV-calculator guard rail that flags any per-tier modeled count exceeding the published total.

4. April 16, 2026v0.3

   Iowa EV calculation switched from sparse-tier-only computation to a default 65% modeled lower-tier payout. The IA path now produces daily EV/Dollar that responds to top-prize depletion without fabricating tier-level precision the lottery doesn't publish.

5. April 16, 2026v0.2

   Indiana scraper migrated to the bucket-modeling path used by the EV calculator's per-state branches. Reconciled IN tier-level counts against the lottery's published top-prize totals.

6. March 17, 2026v0.1

   Initial canonical EV/Dollar formula stabilized across all states: tickets_remaining = remaining_winners × overall_odds; EV = sum(prize × remaining) / tickets_remaining. State-specific data gaps are handled in each state's scraper or in an isolated per-state branch — never by editing the global formula.

All site-wide changes (coverage, pricing, compliance, product) are tracked separately on the changelog.