DavidAnderle

Research software / Working prototype

Volatility Cascade Engine

An independent Python research-software prototype for examining a margin-driven liquidation feedback mechanism in a small synthetic financial system.

Status
Completed
Evidence basis
Supporting artifacts
Started
2026-03
Context
Independent technical and modelling work outside a production trading system.

Role

Independent designer and implementer

Defined the mechanism, implemented and documented the Python baseline, and created 15 synthetic scenarios across six funds and five assets.

Four-step synthetic liquidation feedback mechanism
Schematic of the synthetic feedback mechanism; not an empirical result.

Question

What happens inside a deliberately small synthetic system when losses tighten margin constraints, forced sales depress prices and those price changes feed back into the next state?

Method

The prototype represents six synthetic funds and five synthetic assets. Fifteen deterministic scenarios vary initial shocks and constraint settings. The implementation records state transitions so assumptions and intermediate values can be inspected rather than hidden behind a single result. The current public framing follows the corrected accounting and convergence line of work rather than the superseded early baseline.

What the artifact shows

The prototype demonstrates that the defined mechanism can be implemented and traced consistently under its own assumptions. It does not demonstrate that the mechanism estimates, predicts or explains any specific real market episode.

Evidence maturity

The public project record bundles the corrected 0.3.0-rc1 model specification, reference result, complete 15-scenario sweep and original artifact checksums. This makes the mechanism inspectable as a first-party release-candidate artifact set; independent human clean-room reproduction and accounting review remain pending.

What would come next

Calibration against market data, benchmark comparisons and explicit failure criteria would be needed before assessing whether the mechanism is useful beyond this synthetic setting. That validation is not established by the current prototype.

Mechanism sequence

Four-step feedback mechanism.

Follow the synthetic loss, constraint, forced-sale and price-impact sequence.

  1. Initial loss

    A scenario applies a defined synthetic shock to selected assets.

  2. Constraint pressure

    Fund state is recomputed against the modelled margin rule.

  3. Forced sales

    Positions are reduced according to the explicit liquidation rule.

  4. Price impact

    Synthetic prices update and feed the next deterministic state.

Static sequence: all steps visible.

Claim boundary

What this does not establish

  • The model is synthetic and is not calibrated to market data.
  • It is non-predictive and does not establish real-world causal effects.
  • The corrected prototype is not a trading platform or production risk engine; superseded early-baseline results are not current evidence.
  • The 0.3.0-rc1 evidence remains a release candidate; independent human clean-room reproduction and accounting review are pending, and AI-assisted implementation is not independent research review.
  • Source code is not currently published as a public repository.