Asset management
Crowding and unwind dynamics in a factor trade
Factor relationships hold until the population of participants changes. Simulating the participants makes regime change a mechanism rather than a break in a backtest.
Illustrative example
The question
A factor has performed consistently. How much of that is the factor, and how much is the current population of participants holding it — and what does an unwind look like if that population changes?
Approach
01 · Specify
Participant types with distinct horizons, leverage limits, risk budgets and redemption sensitivities, plus a price-impact rule linking their trading to the price they trade into.
02 · Calibrate
Participation and turnover parameters fitted so that simulated aggregate behaviour reproduces observed volatility and autocorrelation structure, with the accepted parameter region reported rather than a single point.
03 · Validate
Out-of-regime testing against periods the model was not fitted on, and a falsification test stating in advance what behaviour would have invalidated the mechanism.
Output
- Crowding measured as a simulated population state rather than a proxy
- Unwind trajectories under a range of redemption assumptions
- An explicit capacity boundary: the participation level at which the mechanism changes character
Illustrative application. No client engagement, fund or performance figure is described.