Polymorphic Case study
← All case studies

02 / GAMES

The feedback loop that keeps a player economy alive.

We built a mathematical model of a decentralised game economy—tracking sources, sinks, prices, and player decisions—and used it to identify the feedback loop that determines whether the world remains rich with viable strategies or collapses into a single dominant strategy.

The plateTriangle: each point represents a state of the economy, parameterized by the shares of three strategies. Contours: the potential of the underlying game; its basins correspond to stable economies. Shards: populations of players flowing downhill. The model tunes the sink as it drifts toward a fold; without tuning, the economy collapses into a single corner.

ClientINDEPENDENT GAME STUDIO
CapabilityECONOMY MODELLING
Duration14 WEEKS
StatusDELIVERED

01 / the question
What keeps an economy worth playing?

Every action moved the economy.

In a decentralised economy, no central authority sets the rules. Every craft, trade, and reward is a player decision that creates or destroys value, and each decision changes what the next player will find worth doing. Balancing the economy by spreadsheet had produced a familiar cycle: a dominant strategy emerges, gets patched, and is replaced by a new dominant strategy.

Before tuning the system, the studio needed to understand it: which feedback loops kept many strategies viable, and which quietly collapsed them into a single dominant strategy?

Fig. 01

Strategy share in the closed beta

One strategy accounted for 61% of play; the other five split the remainder.

One strategy dominant

02 / the build
Model → simulate → measure → tune

Write the economy as a dynamical system.

We modeled resources, sinks, prices, and rewards as a single coupled dynamical system, with player populations whose choices respond to the payoffs of different mechanics. We ran the model over millions of simulated player-days, calibrated against telemetry from the closed beta.

We measured the richness of the economy directly—how many distinct strategies remained viable—and traced it to the underlying feedback loop: player actions determine resource abundance, abundance determines prices and rewards, and rewards determine which mechanics players adopt next.

  1. Stage 0101

    Model

    Sources, sinks, prices, and rewards as a single coupled system, with player behavior that responds to payoff.

  2. Stage 0202

    Simulate

    Millions of player-days simulated across different populations, including bots, hoarders, and new arrivals.

  3. Stage 0303

    Measure

    Richness: how many strategies remain viable, and which feedback loop is driving it.

  4. Stage 0404

    Tune

    Parameter bands that keep the economy in its rich regime, with early warning signs when it drifts toward collapse.

03 / the result
More ways to play

Balance by mechanism, not by patch.

Before launch, the design team could see which parameter changes would collapse the economy into a single dominant strategy and which would keep it rich. Tuning shifted from reactive patches to a monitored band, with an early signal whenever live play drifted toward the edge of that regime.

2.3×

viable strategies at launch, vs the closed beta

−70%

emergency balance patches in the first season

40M

simulated player-days