Question 1

Where do the EV numbers come from?

Accepted Answer

Pack EV is only as good as the inputs. We treat the board as **research framing**, not tips: cost in versus what you can actually recover, under stated odds and reference prices, after known fees where we model them.

Useful research usually mixes three public layers: (1) **published odds / buybacks**, (2) **pull or pack-open feeds** for a tracked window, and (3) **on-chain receipts** where a chain exposes mints or opens. When a platform publishes nothing, rough odds are reverse-engineered from large samples. When it publishes tables, the interesting work is checking realized hit rates against the brochure.

Always verify live odds on the platform before acting. For the math vocabulary, see [How pack-rip EV actually works](/blog/ev/how-pack-rip-ev-actually-works/).

Question 2

What does “realized favorability” mean?

Accepted Answer

Realized favorability summarizes how packs actually paid out in a measured window versus cost — not a promise of future results. A **negative average** across tracked sites is the normal baseline. “Best” in this category usually means more transparent, more verifiable, and a smaller measured house edge — **not** positive EV.

Short samples diverge from long-run expectation; odds and prices can change without fanfare. Use each logo card on this page to open a dossier, then cross-check [Companies](/companies/) and [Analytics](/analytics/) when volume series exist.

Question 3

How should I rank Pokémon card gacha sites?

Accepted Answer

If “best” means “makes you money on average,” no tracked pack site wins in public mirrors. A useful ranking asks instead: Do they publish odds (and buybacks)? Can pulls be verified (on-chain or high-coverage feed)? How large is the measured house edge? Is the sample big enough?

Start with the platform board above, then deepen with [Why favorability dashboards matter](/blog/ev/why-favorability-dashboards-matter/) and [Net EV after fees](/blog/ev/net-ev-after-fees-platform-comparison/).

Question 4

What’s the difference between published and realized mystery-pack odds?

Accepted Answer

You will meet three kinds of “odds”: **marketing** (vibes, no audit trail), **published tables** (fixed per-tier probabilities, sometimes with buyback blurbs), and **realized rates** (what actually dropped in a measured sample). EV literacy is the gap between published tables and realized rates — and the humility to ignore marketing odds.

Prefer larger sample sizes and fresher timestamps when a “best pack” signal looks exciting. See also [Chase skew and ruin risk](/blog/ev/chase-skew-and-ruin-risk/).

Question 5

How do I compare packs by budget band?

Accepted Answer

Price-band framing answers a narrower question than “which site is best?”: within a spend range, which tracked packs show the least-bad measured cash-back or favorability in a recent window? Even the “leader” in a band is typically still **−EV**. Confidence intervals matter — a hot 2,000-pull sample can look friendlier than a cold 50,000-pull one.

Pick a budget mentally, open matching dossiers on this board, then adjust for buybacks, fees, and exit liquidity ([Net EV after fees](/blog/ev/net-ev-after-fees-platform-comparison/)). Discount unverified / off-chain feeds even when a headline cash-back number looks strong.

Question 6

Why do favorability boards go stale?

Accepted Answer

Published odds are a promise operators can edit. Material changes include edits to a fixed odds table, pack price moves, buyback / cash-back rate changes, and machine rotation that invalidates an old sample. Without a change log, yesterday’s “audited” pack can be a different product today.

Read dossiers with a freshness mindset: useful for comparison, not a live trading terminal. Related: [Why favorability dashboards matter](/blog/ev/why-favorability-dashboards-matter/).

Question 7

What is an odds report / audit?

Accepted Answer

An odds report asks whether each rarity tier’s realized hit rate matches the published rate within sampling noise. Platforms that publish per-pack odds are auditable this way. Platforms that publish nothing force reverse-engineering from pulls — informative, but a different confidence grade.

Prefer dossiers that show sample size and caveats, and use [Analytics](/analytics/) for volume context, not just pull EV.

Beezie EV

Beezie · EV research framing

Other platforms