Marvel Snap SC Score: Tier List Ratings Guide
The SnapComplete Score (SC Score for short) is our single-number rating for every card in Marvel Snap. It folds cube rate, win rate, and play rate into one value between 0 and 100, adjusts for cards that only work next to one partner, then drops each card into a tier from SSS down to F. One glance tells you how strong a card is right now.
Why a Single Number?
Marvel Snap gives you three obvious performance signals: cube rate (how many cubes a card nets when drawn), win rate (how often it wins when drawn), and play rate (how often people slot it into their decks). Cube rate rewards smart snap-and-retreat play, win rate rewards consistency, and play rate reflects meta presence and deckbuilding flexibility.
No single signal tells the whole story. A card can post a 55% win rate but a negative cube rate because it wins small and loses big. Another can post a great cube rate yet appear in only 2% of decks because it needs a very specific build. SC Score rolls all three into a weighted composite, so you don't have to triangulate three leaderboards in your head.
How the Score Is Built
We convert each card's raw cube rate, win rate, and play rate into a percentile rank across the full card pool. A card at the 90th percentile of cube rate outperforms 90% of other cards on that signal. We then weight the three percentiles, with cube rate carrying the most weight because it captures the snap-and-retreat dynamic that wins and losses alone ignore. The weighted sum is the 0 to 100 SC Score.
Low-sample cards get Bayesian shrinkage. If a niche card has only 50 matches, we pull its raw numbers toward the pool average so one lucky streak can't inflate its score. Every card that appears in the window gets a score, and cards with tiny samples land near the middle tiers because the shrinkage says "we don't have enough evidence to call this one strong or weak yet." As matches accumulate, scores drift in whichever direction the real signal points. The shrinkage strength is calibrated from our live data, so it auto-tunes as the match pool grows.
Update, September 24, 2026: We now account for synergy. When a card is mostly good because of one partner, part of its results really belong to that partner, so it gives up a few points. The card a deck is built around keeps its full score. That stops one dominant deck from pushing its whole list to the top. See Relies On below for how a card qualifies.
The Tier System
Tiers use absolute score bands, so a score of 75 always means A tier, no matter the window. The elite range (90+) gets one extra wrinkle, cliff detection: when the top cards leave a visible gap between them, the boundary snaps to that gap so a truly isolated card earns SSS alone.
| Tier | Score Range | Read |
|---|---|---|
| SSS | 97+ (or above the top cliff) | Broken, often nerfed next |
| SS | 93 to 96 | Meta-defining |
| S | 90 to 92 | Strong, widely played |
| A | 75 to 89 | Good role-player |
| B | 60 to 74 | Solid mid-tier pick |
| C | 40 to 59 | Fringe, needs a specific shell |
| D | 20 to 39 | Weak, rarely worth the slot |
| F | Below 20 | Buff candidate |
SSS, SS, and S tier letters show in cyan on card rows. A through F render in plain text.
This OTA vs Last OTA
Marvel Snap ships balance patches (OTAs) roughly every two weeks, and each one reshuffles the meta. So we track two separate windows:
- This OTA is the live window since the most recent balance patch, recomputed every hour as new matches stream in.
- Last OTA is the previous window, frozen at the moment the new patch dropped. It never changes again.
Right after an OTA drops, This OTA has barely any data, so the shrinkage prior drives most of every score (cube and win rate pulled toward pool average). Tiers still populate across the full SSS-to-F range, but they carry less signal in the first day. Want a more confident read? Switch to Last OTA (Premium) for the stable 2-week pre-patch rankings. The live scores sharpen on their own as post-patch matches accumulate.
What the Score Is Not
SC Score is a per-card rating, not a deck rating. A card can be SSS and still lose you games if you build the wrong deck around it. Use the Decks page to find the lists that actually win.
It's also not a verdict on whether to play or cut a card. Tech picks, personal playstyle, and bracket optimization all matter, and the score can't see any of them. What it sees is aggregate Ranked performance from tracked matches played at Collection Level 16,000 and above. Treat it as a starting point for research, not the final word.
Relies On
Some cards are good partly because of another card. If a card shows up with a partner more often than the partner shows up with it, wins more when that partner is in the deck, and gains more from the partner than the partner gains from it, its page shows Relies on with that partner and its win rate with and without it. Those are also the cards that lose a few SC Score points. The more of its games depend on the partner, and the more lopsided the pairing, the more they lose, so a partner that's only in some of a card's decks costs a point or so.
What Data the Score Uses
SC Score reads from tracked Ranked and Ranked Infinite matches on SnapComplete, minus bot matches, played by people at Collection Level 16,000 and above. Those players own most of the card pool, so the score measures whether a card is worth chasing rather than how it performs against collections still missing the alternatives. Conquest, LTGMs, and Battle Mode don't count toward it. The window is always pinned to the most recent balance patch, so the numbers reflect the game's current balance state. If Second Dinner rebalances a card midway through a window, that card's score moves as fresh post-patch data lands.
The pool is every released card. Cards that show up in decks but were never drawn in a match still get a score, with Bayesian shrinkage parking them near the pool average. Low-sample cards aren't rejected; they're scored with the uncertainty built in.
Why the Secret Sauce Stays Private
The exact weighting ratios stay internal. We publish the methodology so you can understand what the number means, but not the knobs that set it, because the score is meant to reflect how we read the meta, not how anyone reverse-engineering a formula would. Anyone who wants the underlying raw cube, win, and play rates already has them: they're on the Meta tab under other sort options.