Roundel.
If a system looks stable and models assign reassuring probabilities, Roundel maps where resilience breaks under stress, without claiming to know the black swan.
Fragility map & reverse stress testing
Ruin events are too rare for a precise probability. Roundel applies stress multipliers (2×, 5×, 10× historical volatility) to exposures known today — energy costs, leverage, loss of a key supplier — and watches where the response stops being linear and harm grows disproportionately.
That breaking point is the core output. With reverse stress testing, Roundel also starts from a declared ruin scenario («what would need to happen for this system to collapse?») and measures today’s safety margin, including for scenarios absent from historical data.
Recommendations remove fragility: single points of failure, excess leverage, excess concentration. They favour cutting structural exposure over betting on a forecast of risk.
Roundel principles
On ruin, ordinary calibration says little. What counts are contestable stress tests and a record of real shocks weathered without collapse.
Non-linear stress multipliers
Applies growing stress to current exposures to see where impact scales disproportionately relative to the shock.
Reverse stress testing
Defines the total-ruin threshold and calculates how far, from the present state, the system is from reaching it.
Interventions by subtraction
Identifies dependencies to remove: vendor concentration, contractual rigidity, exposed leverage. Aim: more resilience, less added complexity.
How it fits together
Vektor works on ordinary risk. Probe on agents’ futures. Roundel on fragility under extreme stress, when a reassuring output is no longer enough.