Marginal card power from an equal-exposure simulation: every card gets ~equal games on random teams versus random opponents. Because both decks draw the same rarity mix, the metric is normalized within each rarity β 0.500 is the average card of its own class, so these rank cards against their peers (captain vs captain, epic vs epic), not across rarities. Run under two pilots to bracket play style, then averaged.
Yes β but it's entirely the ability. Isolate ne correctly (neutralize only its kit while every other captain keeps theirs) and its raw card is below average β .481. Chain Reaction alone accounts for the whole #1 standing: +.127 win rate, by a distance the strongest ability in the game (a good ability like Adaptive is ~+.05; VulcanForge's whole kit moves Β±.02).
The mechanism is the whole-field stamina refund. It's the only ability that injects stamina, and it hits all four fielded cards (+1.5 each per chem convert, ~3.5 fires a game β 18 stamina β a fifth of a legendary's budget, re-upped every game). ne's entire bench stays fresh and out-buzzes everyone late, so its 87/87 phys & math win the non-chem rounds too. Turn the refund off and ne is bottom-half. So the nerf lever is the refund (its size, or refunding only the converting card) β not the stats, and not the chem-only trigger.
One caveat the sim can't show: chemistry is ne's only trigger and its best stat, so a human can deny it. Emulated, starving ne of chem drops it .62 β .59 (half denied) β .51 (fully denied). Real-play ne likely sits ~.55β.60 β still top-tier, just not the .61 the (denial-free) sim shows.
A separate sim: seat A drafts a fixed captain + epic (rest random), seat B is fully random. Win rate is the pair's absolute strength vs an average opponent (.500 = a coin flip). Synergy = how much the pair beats the sum of the two cards played apart β positive means the two are worth more together (shared real team/season trait, or complementary category coverage).
ne + AdiyaSivahuma is the outright best pair (.750 β wins 3 of 4 vs a random deck): the two strongest cards, both NSBA1, so a season trait on top. ne anchors 11 of the top 15 duos. If you can't get ne, akhil + Daniel S (.652) is the best captain-epic combo. On pure synergy, the standout is the one legendary+epic pair that shares a real team β Rohan G + Atlasguy101 (both 2y4b), +.082 over their solo value: the narrow team trait genuinely pays, even though both are mid cards on their own.
The best cards within rare, uncommon and common β the ones to reach for when a slot is cheap. These win rates are normalized inside each class, so they rank a card against its own rarity; they are not power-comparable to the epics above (that needs a different sim β see the note at the bottom). What they share is the same DNA as the top epics: a single deep spike. Earth/Space one-tricks are conspicuously over-represented here.
Spike > breadth. The board self-sorts by category, so a fielded card only ever answers its strengths and its weak categories barely cost it. A 92-in-one-thing specialist wins its category outright; a flat 78-across generalist wins nothing outright. Across the pool, spread (best minus worst category) predicts win rate at r=+.56 while average knowledge predicts nothing (rβ0). That is why JoshuaW (spread 17) is the worst epic and the dual-spike Daniel S (phys 91 / math 89) tops it.
Posture is a second axis. Under an aggressive pilot (contest the buzz every question) broad cards gain and fast cards fade. sanj is the extreme: the worst captain on autopilot (.380) becomes a B when piloted hard (.518, a +.139 swing) β his breadth only pays if you press it. He is a skill-expression pick, mis-rated by any single number. akhil swings the other way (AβD under aggression); yufei and Jianyu rise with it.
ne is a real outlier. Triple hard-science peak (chem 92 / phys 87 / math 87) plus Chain Reaction refunding stamina on chem converts β it leads under both pilots and sits +.06 clear of the field. Every other captain ability barely moves win rate (their points-per-tossup β non-captain baseline; only ne's β 2Γ).