A production-grade execution kernel for stress testing that is replayable, attributable, and debuggable.
Coromandus builds PlanckNet: a deterministic stress-testing kernel that decomposes portfolio losses into market moves, liquidity impact, and control interventions, so risk teams can identify what fails first under stress and validate guards before crises.
Designed for investor and risk-team evaluation: deterministic replay, decomposed attribution, and API-first integration.
The problem
Institutions often cannot explain why portfolios break under stress. Standard tools collapse complex failures into black-box metrics.
Black-box risk metrics
Aggregated metrics (VaR, volatility) hide failure modes. When something breaks, teams cannot isolate what actually failed.
Liquidity blindness
Many models assume you can always trade. Execution costs during stress are treated as crude add-ons or ignored.
Non-reproducible tests
Stress tests are often one-off reports, not repeatable experiments. If the crash cannot be replayed, it cannot be debugged.
Our solution
Deterministic, decomposed stress experiments. Same seed, same crash path, every time, with attribution that shows why.
Stress testing as an experiment
Change regimes, inject shocks, validate controls. Treat stress runs like a lab: reproducible, falsifiable, and debuggable.
- Controlled regime transitions (volatility and liquidity)
- Deterministic replay for governance and auditability
- Step-by-step attribution for decision-grade insights
From "we lost" to "we know why"
Every timestep separates the components that get mixed together in standard workflows: market movement PnL, liquidity shock cost, and guard intervention.
- Market PnL: price moves on existing positions
- LSO cost: execution and impact under depth and urgency
- Guards: testable circuit breakers with measurable avoided loss
Core technology
PlanckNet is built around three operators plus a deterministic kernel layer, designed for multi-portfolio stress simulation.
RPO: Risk Propagation Operator
Models how shocks propagate through correlated instruments, enabling scenario-driven factor stress.
LSO: Liquidity Shock Operator
Quantifies execution cost based on depth and urgency, making liquidity impact first-class rather than an add-on.
Drift Guard System
Per-portfolio circuit breakers with rolling budgets, measurable guard activation, and avoided-loss reporting.
Determinism by design
Seeded execution ensures identical runs produce identical outcomes. This enables real debugging, governance review, and stable evaluation of guard policies.
API-first integration
Expose the kernel via REST to run in your environment and return structured artifacts that can plug into existing risk workflows.
Live demo
The demo below links to your hosted console. Use it as the fastest investor proof: same seed, same crash, fully inspectable output.
What to show in 2 minutes
Investors should see one thing: a debuggable crash, not a feature tour.
- Run a conservative preset and a fixed seed
- Point to the step where volatility and liquidity flip regimes
- Show decomposition: market vs liquidity vs guard impact
- Re-run same seed to prove determinism and replayability
- Switch preset to show failure mode changes
Tip: If the iframe is blocked by browser policy, use the button to open the demo directly.
Business model
Start with hypothesis-driven paid pilots, convert to annual licenses, expand across desks and strategies.
Paid Pilot
$5K to $10K
- 2 to 3 falsifiable hypotheses
- Reproducible stress reports
- IC-ready walkthrough
- Calibration and scenario pack
Annual License
$100K to $150K
- API + console access
- Deterministic replay as workflow
- Multi-portfolio strategy comparison
- Guard validation and reporting
Expansion
$200K+
- VPC or on-prem deployment roadmap
- Audit logs and structured artifacts
- Multi-desk expansion
- Scenario packs and governance support
Target ICP
Hedge funds and asset managers first, then banks. Initial buyers are quant research and risk innovation teams.
Sales cycle
Typically 3 to 6 months for hedge funds, 6 to 9 months for asset managers, longer for banks.
Why now
Liquidity risk and governance demands increased after recent crises, pushing explainability and reproducibility to the front.
Status
This is a running engine, not a mockup: live API, console, and a formal specification behind the kernel.
What exists today
- Production-grade simulation kernel (V18.3.4.2)
- Live API + console demo
- Formal spec (1,700+ lines) behind execution logic
- Multi-portfolio support with three archetype strategies
- Deterministic reproducibility across runs
What we validate in pilots
- Liquidity attribution under stress (how much of tail loss is execution cost)
- Guard effectiveness (avoided loss) and threshold selection
- Portfolio failure mode mapping under controlled regime changes
- Governance-grade replay: same seed, same crash path
The ask
Pre-seed capital to harden the API, complete validation work, and convert initial pilots into recurring licenses.
Raise details
- $200,000 pre-seed
- Instrument: SAFE
- Valuation cap: $2.5M to $3.0M
- Discount: 20%
- Runway: 9 months
- Milestone: 2 to 3 paid pilots, Seed-ready
What an investor should remember
PlanckNet is the experimental layer alongside existing suites: deterministic replay + decomposition + guard validation.
- Deterministic execution is hard to retrofit into stochastic engines
- Decomposition turns stress tests into actionable engineering outputs
- Moat compounds through pilots: calibrated scenario packs and institutional learnings
If you want a contact section with email, replace the placeholder below in the footer.
Next step
Fastest path: open the live demo, run a fixed seed twice, and evaluate attribution plus guard behavior.
Book a demo
Use the live console to validate the core claim: same seed yields the same crash path, with stepwise PnL decomposition.
Investor materials
If you are hosting additional PDFs, link them here (investment memo, deck, one-pager). Replace the href attributes with your hosted files.
Optional: add analytics, form endpoint, and investor PDF links after deployment. Current page has no trackers and no external forms by default.