Case study 02 · Systems modelling

Agent-Based
Market Simulation

A behavioural simulation that asks how changing market participation and strategy shifts can produce boom-and-bust cycles without an external shock.

NetLogo
Agent-based modelling
Monte Carlo simulation
Genetic algorithm
Python
Empirical validation

Model

Three behaviours, one evolving market.

The model introduces savers alongside fundamental and technical traders. Savers participate when market strategies outperform saving; traders can switch strategy through relative performance and social interaction. Their changing mix feeds back into price formation.

A representative agent-based simulation showing log price, returns, market flow and the changing composition of agent types. — A representative run: price, returns, market inflow and outflow, and the evolving distribution of agent strategies.

Result

Participation dynamics create regime shifts.

Inflows are endogenous. Strategy performance can draw savers into the market and amplify prevailing dynamics.
FOMO raises volatility. Restricting new entrants to trend-following behaviour produces stronger price variability.
The model is empirically grounded. Distributional and temporal properties are compared against the Bitcoin analysis.

Comparison of simulated and empirical distributions, tail indices and return autocorrelations. — Simulation runs compared with empirical distribution, tail and autocorrelation patterns.

Comparison of unrestricted and FOMO-like trend-following scenarios. — Constraining new entrants to trend-following behaviour increases return volatility.

Interactive model

Explore the system directly.

Adjust a small set of meaningful parameters in a dedicated NetLogo Web simulation and observe how the agent population, market flow and price path respond.

Open the simulation

Context: Independently authored research and engineering project, developed as a B.Sc. Economics thesis at Leipzig University.