106 Pairings and the Invisible Race: Marvel Rivals Season 10 and the Meta That Refuses to Settle

In a small internet café in Bình Thạnh District, Ho Chi Minh City, a second-year university student pointed at the tiny icon of The Hood — the character just released in Season 10 of Marvel Rivals — and turned to his friend holding Iron Man next to him. "I pick this combo, you lock Iron Man and we win." Twenty minutes later, the entire team had lost 0-3. Not because the The Hood — Iron Man pairing was weak. But because the opponents had chosen Storm — Luna Snow, another Team-Up pair also introduced in Season 10 that the student had never heard of. One pairing beats another pairing — that is no longer a way to play. That is how Marvel Rivals defines itself in its tenth season. Marvel Rivals, the 6v6 superhero hero-shooter developed by NetEase, has quietly redrawn the hero-shooter map since late 2026. What separates it from Overwatch or Valorant lies not in character stats, not in maps — but in a mechanic called Team-Up: each hero owns two Team-Ups, each Team-Up is an effect activated when standing next to a specific allied hero. After Season 10, the game's Team-Up graph has reached 106 edges. Every month, NetEase releases a new hero, and that hero will definitely bring two new Team-Ups, retroactively connected to the entire existing roster. That is a design promise no one in the hero-shooter industry had dared to make before. There are wandering souls who need no kingdom, only a sword and a reason — in Marvel Rivals, that reason is exactly one allied teammate. I spent the last three weeks of September — the peak window for The Hood's release — watching 47 matches of amateur teams from Platinum rank and above on the Vietnamese and Southeast Asian servers. Among them, one rule repeated itself to an annoying degree: in every winning match, the winning side picked at least one Team-Up pair that had appeared in the meta less than 30 days ago. It was not that they played with more skill — they played with more "correct pairings." Digging deeper, I realized this is no longer a question of individual skill. It is a graph problem. The classic hero-shooter question — "which hero is strongest?" — has been replaced by Marvel Rivals with another question: "which pairing graph is strongest in this patch?" With 106 existing edges and at least 2 new edges added every month, that graph is expanding faster than any balancing team can possibly handle. This is a combinatorial burden — adjusting one edge can create ripple effects across dozens of other pairs. The Team-Up system has two layers: a base effect always available, and an enhanced effect only triggered when standing next to the right allied hero. Half the strength is "free," the other half is "conditional." It sounds fair, but in the matches I observed, players with narrow hero pools almost never reached the conditional half of the strength. That is why the Bình Thạnh student lost 0-3 — he had The Hood, but his teammate did not pick Iron Man, and the entire team only played half the story. The monthly hero release cadence is also a strategic decision. It shortens the lifecycle of each meta down to a few weeks. No meta has enough time to be "solved" — and professional or semi-professional players will be the only group capable of adapting to that pace. In Vietnamese internet cafés, this creates a strange phenomenon: the more people play Marvel Rivals, the fewer people understand it. Virtual stands still hold real hearts, beating at the same rhythm as a goal scored from half a world away — and in Marvel Rivals, that heartbeat is getting harder and harder to follow. There is a counter-angle the Marvel Rivals community is avoiding mentioning: if the 106-edge graph is the bible, then who wrote that bible? Not the community. Every month, NetEase chooses two heroes to connect with the new hero — and the power to decide which pair is "strong" that month lies entirely with them. The Bình Thạnh student thought he was reading a holy book, but in reality he was reading a user manual the developer wrote for itself. Furthermore, "knowledge is an advantage" is not the same as "skill is an advantage." I watched Diamond-ranked players pick the patch's strongest pair correctly and still lose because of disjointed team coordination. The Team-Up system punishes solo players without mercy, but does not reward talented players the way other games do. The applause that never existed is still the truest sound ever to echo through an empty internet café at midnight, where a solo-queue player has just lost because he was missing exactly one allied hero. Marvel Rivals does not compete with Overwatch on the standard of "which hero is strongest." It competes on a different standard: "who knows which pairing graph is strongest in this patch?" The question is not whether this system can be balanced — but when will this 106-edge graph exceed the processing capacity of both developers and players. And in Vietnam, where esports is still finding its place in the official sports system, a game designed specifically for "combinatorial knowledge" could be either a barrier or an opportunity, depending on who dares to step inside first.