Marvel Rivals and 106 Team-Ups: When Game Balance Outruns the People Who Built It

In a small apartment in Busan, I rewound the final teamfight of a ranked match and paused on the draft screen. It was not a clutch play that made me hit pause, but two hero slots locked next to each other — a Tank and a DPS sharing a Team-Up — and the way the opposing team completely misread that intent. Twelve minutes later, the match was decided. That was the moment I understood Marvel Rivals was no longer asking "which hero is strongest." It had become "which pairing is strongest under the current patch."

With 106 Team-Ups in existence, two Team-Ups per hero, and a new hero arriving roughly every month, NetEase's 6v6 shooter has accidentally built a synergy web so large that nobody — not even their own balance team — can track it fully in real time.

Let me be clear from the start: this is not a celebration of a game. This is a math problem.

Context — a mechanic that defines the meta, not heroes

Marvel Rivals is a 6v6 hero shooter where Team-Up serves as the core differentiator against every competitor in the genre. Every hero has two Team-Ups. No character is released without one — that is a design commitment, not an improvisation. Each Team-Up has two parts: a base effect that is always available, and an enhanced effect that only triggers when the corresponding partner appears on the team. Season 10 brought The Hood into this system, adding new pairings. And most importantly: the developers confirmed new heroes will pair with old ones rather than replace them.

Based on my experience following matches from the closed beta to now, this is the life-or-death difference. In other hero shooters, power is measured by whether a hero performs its role well. In Marvel Rivals, power is measured by a connectivity graph. A hero can be average in isolation but becomes a mandatory pick when its Team-Up partner is locked in alongside.

This turns the draft phase from an individual decision into a combination decision. And combinations have no ceiling.

Core — a combinatorial problem with no tidy answer

Let us do simple math. If each hero has two Team-Ups and the roster grows by one unit each month, then each new hero does not merely add two edges to the graph — it also raises the power ceiling of the old heroes it pairs with. This ripple effect is not linear. It scales.

In other words, the balance workload NetEase must handle exceeds the speed at which they can tune it.

When I was a social media commentator, I saw something similar in a turn-based strategy game: the more new units were added, the harder it became to guarantee no single combo dominated. But there, win-rate data per pairing existed. Here, that data is not published. We have an inventory — how many Team-Ups exist, which pairs they are — but not a single performance number: no win rate, no pick rate, no ban rate.

That means every judgment about the current meta direction is structural, not data-backed. And the structure is saying something fairly clear: the system is pushing players from choosing the strongest hero to choosing the strongest web.

Who benefits? Flexible players with deep hero pools, who read the Team-Up map before locking in. Who loses? One-trick players, lineups built around a single strong individual, and especially solo ranked — where coordination is not guaranteed. The enhanced effect sits behind a conditional door: you must have the right partner. In a random environment, that door is usually shut.

This is the point I find most notable. The mechanic rewards organized preparation and punishes improvisation. In design terms, it creates a high and rising knowledge barrier. 106 Team-Ups exceeds the reactive memory capacity of an ordinary player. The author of the original guide even advised readers to bookmark the page for frequent reference. That is a clear signal: newcomers will be overwhelmed, veterans hold accumulated advantage, and professionally coached players hold the biggest advantage of all.

In long-term design terms, the commitment that "new heroes always pair with old ones" is a wise choice. It protects the value of older heroes and avoids obsolescence through power creep — the kind of mistake that makes many seasonal live-service games quietly destroy their own rosters. But the price is a permanently growing synergy web, and therefore a linearly increasing balance-testing load.

Contrarian — where I could be wrong

I must be honest: the entire argument above rests on a structural foundation, not a performance one. I do not have the win rate of any Team-Up pairing. No per-patch power rankings. No data on The Hood's appearance frequency after Season 10 launched.

The mocked man often holds the real data, but this time the mocked man is me — because I am judging without a single number in hand.

There are three ways I could be wrong. First, the always-available base effect may be softening the coordination pressure more than I think. If a hero retains baseline value without its partner, then "forced pairing" is only half the story. This is described by the source but I have not independently verified it — it needs confirmation.

Second, a large web is not necessarily a chaotic one. If the balance team distributes power sensibly, sheer quantity can act as a shield against any single combo dominating — because there are too many alternatives to counter it. Fighting-game history shows this is sometimes true.

Third, a monthly hero cadence may not be a burden but a retention weapon. It ensures no meta is ever "solved," and for a seasonal live-service game, instability is often growth fuel rather than a sign of decline.

I hold my position, but I mark the unverified zone clearly. A provocative writer without evidence is just a cheap scold. I do not want to fall into that.

One small but notable detail: the original guide carries a September 14 update stamp but no year. For a document committed to self-updating, the missing year weakens its authority as a "current" source. A figure like 106 Team-Ups can silently drift out of date between updates. That is a small but real data risk.

Fans worship legends, but forget that legends only survive by being verified.

What I predict

If the Team-Up system keeps expanding at one hero per month, I expect that within a few seasons the community will generate its own "blacklist" of mandatory bans — not because the developers want it, but because the graph structure encourages it. When the web is dense enough, players will always find the optimal pairing, and at that point the draft becomes a war between combinations rather than between individuals.

A team does not collapse on the night of destiny; it rots long before in silence. And a balance system does not break in a single patch; it slowly overloads, edge by edge.

The question I leave to those steering this game: if every new hero must add two more edges to an already dense web, then by month twenty, who will be the one to check the two-hundred-and-sixth edge?

And for players, this is a simple reminder: look it up before locking in your hero. Not to copy, but to know what you are leaving on the table.