Methodology
How we source, calculate, and — where we can — verify every number on RipIndex. No black box: here's exactly what each figure means.
Where the numbers come from
Every figure comes from public data — the odds each platform publishes, the pack-opens they record, and, for on-chain platforms, the blockchain itself. We don't take a platform's word for anything we can check independently. Data can lag or be incomplete, so we timestamp freshness on each machine and recommend verifying current odds on the platform before you act.
How we calculate EV
We measure a pack's value the way that's actually guaranteed to you: cost in, versus the buyback value of the card you pull. Most platforms we track offer instant buyback — a standing offer to buy your pulled card back at a set percentage of its market value — making buyback the one return you can realize with no resale, timing, or liquidity risk. Where a site works differently, its section below says so.
EV = the probability-weighted buyback value of the pull, divided by the pack price. We show it as an "edge" (EV minus 1). On every machine we track, that edge is negative — the house edge. Opening packs is negative-expected-value gambling; the math just says how negative.
How we verify — published vs. reality
Published odds are a promise. For on-chain platforms (such as Courtyard on Polygon and Collector Crypt on Solana), we can check whether that promise holds: we read the actual pulls off the blockchain and compare what really landed against what the platform advertised. That is our odds audit — "published X%, actually Y% over N pulls." Most odds trackers compute EV from published numbers and stop there. We check them. For off-chain platforms we can't read a chain, so we track the platform's own recorded pulls and are explicit about the difference.
VGEV — why an average EV can mislead
A pack can show a healthy-looking EV while a typical open returns pennies on the dollar, because the average is propped up by a rare "grail" that almost no one hits. So alongside EV we show:
- Typical open — the median result. Half of opens return less than this.
- Loss chance — the share of opens that come back worth less than the pack cost.
- Volatility — how wildly outcomes swing (Low / Med / High).
- VGEV (volatility-adjusted EV) — the EV docked in proportion to that swing (VGEV = EV minus one-half of the standard deviation of returns), so grail-lottery packs are penalized for the risk their headline EV hides.
Averages describe a crowd; these describe you.
The tier board — how "best pack for your budget" is ranked
The tier board buckets every pack we track into price bands and ranks each band by realized 7-day cash-back — what real recorded opens actually paid, valued at each site's own instant buyback, as a share of pack price. Published or theoretical odds never rank. A pack needs at least 200 measured pulls to compete, and a band winner is only called when the leader's 95% confidence interval clears the runner-up's — overlapping intervals read "too close to call." Packs we can't measure (no instant buyback, or no per-pull feed) are shown but never ranked, with the reason stated.
Neutrality: tier-board rankings are computed from measured pulls only. They are never influenced by referral, partnership, advertising, or any commercial relationship — some "Open" links on that page are referral links, and they have no effect on what ranks where. And the part that never changes: almost every pack on every site is negative-EV by design; "best" is the option that loses the least on average within its band, and a short-window reading above breakeven is sampling luck, not an edge.
How we read each site
Each site exposes different data, so the depth of what we can measure differs. This is exactly what we can and can't verify, site by site. (These notes used to sit on the board itself; they live here now.)
- Collector Crypt (Solana) — realized EV comes from each machine's recent-winners feed, valued at the machine's own instant-buyback percentage. Current odds come from the live weighted insured value of the actual pool. Grail restocks are detected from full-pool snapshots.
- Courtyard (Polygon) — publishes odds, EV and buyback directly, and we audit them: realized EV, per-tier realized odds, and live heat all come from pack-opens read straight off Polygon. Newer and pricier packs' signals grow as pulls accumulate.
- Phygitals (Solana) — publishes odds, EV and buyback directly, and we audit them: realized EV comes from their live pull feed (a sampled stream, so thin windows read "accumulating"), the per-tier audit compares published odds to what actually drops, and pulled cards are verified as real NFTs in their Solana vault. We also maintain a complete on-chain pull ledger: following the machine-vault wallet records every pull with its exact chain timestamp — which is how we measured that the public feed displays only a fraction of pulls, skewed toward big hits. The site-wide census on the board comes from that ledger, valued with Phygitals' own card prices; per-pack realized numbers are still shown only where the feed sample proves representative.
- GachaPull (Abstract) — realized EV from its live pull feed, valued at buyback — the honest cash-back, not the theoretical number.
- Monster (MegaETH) — the headline is real lifetime cash-back from Monster's own exact counters, after the buyback haircut; published tier EV overstates.
- Beezie (Base + Solana) — claw machines with no per-pull feed, so per-open and per-card outcomes aren't observable. But the instant buyback settles in USDC on-chain, so realized recovery IS measured: SITE-WIDE on Base (its machines share one treasury, so per-machine isn't cleanly separable) and per-machine on Solana. No per-card odds audit — so it doesn't rank on the tier board — but the measured cash-back recovery is real.
- Tilt Rips (off-chain) — odds & EV are live and realized EV is forward-only from its public feed, but there is no independent ledger to verify its published odds against. Treat its odds as unverified.
Whatever the depth: almost every pack on every site is negative-EV long-run. Deeper data changes how well we can measure the loss, not whether there is one.
Limits & honesty
- Small samples are noisy — early data on a machine can swing a lot. Heat reads are volume-weighted (a small window is blended toward the machine's own weekly average, so a couple of lucky pulls can't fake a hot streak), and any read under 25 pulls shows as a neutral "thin read" — never hot or warm.
- Published odds are not guarantees, and platforms can change them at any time.
- Past pulls don't change your odds. A machine "running hot" is recent variance, not a prediction.
- It's still gambling. Every machine is negative-EV long-run; none of this is a way to make money, and none of it is financial or gambling advice. Never spend money you can't afford to lose. US help: call or text 1-800-GAMBLER.