Marvel Rivals: 106 Team-Up Combos and the Balance Math Nobody Wants to Count

Hook

At two in the afternoon Seoul time, I sat down and counted the Marvel Rivals Team-Up list by hand. One hundred and six. Not one hundred and six lines of text, but one hundred and six edges connecting heroes inside a synergy graph that keeps swelling every month. I stopped there for a while. The last time I saw a game push its ability-combination count into triple digits was when Overwatch 2 was still fumbling with its role-based balance problem, and even then the number never touched this threshold. What caught my attention was not the glamour of a long list, but the rules standing behind it: every hero has exactly two Team-Ups, no hero ships without one, and every new hero must be able to shake hands with the old ones. As someone who once spent forty-seven days just listening to the ghosts of crowdless passes, I recognized immediately that this is not a content story. This is a governance story.

Context

Marvel Rivals is a 6v6 hero shooter developed by NetEase, and Team-Up is precisely the mechanic that differentiates it from Overwatch. The system runs on two layers: a base effect always available to every hero, and an enhanced effect that only activates when the correct partner hero stands beside you in the lineup. Put another way, half of a pairing's power is free, and half is locked behind a matching condition.

Season 10 just added The Hood along with that character's new Team-Ups. The developer does not hide its ambition: a new hero roughly every month, and every new hero must share screen time with an old one. This is a design commitment that behaves like a long-term contract, not a one-off patch.

To make it intuitive for an outsider: if Overwatch plays at the level of "which hero is strongest," Marvel Rivals plays at the level of "which pairing web is strongest under this patch." The question itself has changed at the root.

Core

The first thing to face squarely: Team-Up is the most meta-defining lever in this game, and it turns the meta from a hero-pick problem into a pair-pick problem.

When you pick a hero, you are not just picking its kit. You are signing an implicit contract with another hero on the draft board. If that partner never shows up, the enhanced layer vanishes and you keep only the base layer — meaning you are playing a shortened version of yourself.

The direct consequence is that balance no longer lives in individual bodies. It lives on the edges. A small change to one pair's enhanced effect can ripple into three or four different compositions at once. With 106 edges and a monthly hero cadence, each new hero does not merely add a character — it adds at least two new edges to the graph. That is the combinatorial balance burden: the more pairings exist, the lower the chance of guaranteeing that no single pair dominates.

The second point: scale has crossed the threshold where encyclopedic knowledge becomes a competitive asset in itself. The original guide itself advises players to bookmark the list for frequent lookups. When a game forces players to open a reference book before locking a hero, you already have a high knowledge barrier, and it is still rising. That barrier favors veterans and coached players, and works against newcomers.

The third point: a monthly release cadence means the meta never fully stabilizes. Adaptation windows get compressed. For a professional scene, that means the window for a team to "solve" a meta is shorter than the window they need to build on solid ground.

The fourth point, and this is what I want to stress: the two-layer design is a deliberate way to relieve pairing pressure. The existence of a base effect means a lone hero still keeps baseline value. Roster pressure is not erased, but it is softened. This is an easily missed detail when people stare only at the flashy enhanced half.

On a professional draft board, the consequences are very concrete. Bans no longer target a strong hero; they target a strong edge. You can neutralize a pair by removing one end. Conversely, a team with a deep hero pool will always have a detour when their favorite edge gets blocked.

Contrarian

Now comes the part where I have to hit myself, as I still do every week.

Everything I just wrote above is structural analysis, not data analysis. The original article only supplies inventory: how many Team-Ups exist, who pairs with whom, how many combos each hero has. It contains not a single performance figure — no win rate, no pick rate, no ban rate. Without data, every judgment about meta direction is inference from design, not evidence.

I have been wrong in this exact way before. Three times misreading Modrić taught me that a match does not need to be read correctly, only deeply. But reading deeply without numbers is still reading deeply in the dark. Numbers say he exists; instinct says why he is terrifying — and here, I only have instinct about structure.

There is one more doubt. The original article's description of how base and enhanced effects split may not match the actual in-game mechanic. If the real split ratio differs, the degree of partner dependency differs too, and my entire conclusion about pairing pressure would need adjusting. I leave that open.

And the number 106 itself needs cross-checking against official patch notes. A frequently updated guide can still drift silently between two edits.

Takeaway

My verifiable prediction: within the next six months, the Marvel Rivals balance team will have to publish at least one adjustment aimed directly at a specific Team-Up pair — not at a single hero. The reason is simple: you cannot manage a 106-edge network that grows by two edges a month just by tweaking individual heroes one by one.

I saw Haaland in the xG pile before the whole world called him a monster. Here, I see no monster at all. I see a net, and a pair of scissors running slower than the net weaves itself.