Tail-Risk Engineering — Amsterdam
Concentration is a position. Hedge it like one.
Your net worth is one ticker. We run 50,000 extreme macro paths against your actual position and hand you the shape of your left tail — priced, hedgeable, with the model's failure rate printed on page one.
3 fields · 48-hour document · no call, no sequence. 1.28 billion paths run to date.
50,000
simulation paths per diagnostic run
0.8 s
median compute time, full re-run
−31.4%
ES97.5, 1987-replay stress, 80% single-stock book
0.4% p.a.
indicative hedge carry at −20% floor
Engine benchmark 2026-05 · illustrative parameters · methodology and failure cases documented below.
You optimized the company. The position that pays for your life is still un-load-tested.
Engineered resilience
“I built this — why would I distrust it?”
Because the system that created your wealth is not the system that protects it. You ship code against test suites; your concentrated equity runs in production with no failure mode coverage. The same engineering rigor you apply to uptime has never been pointed at the one asset holding 60–90% of your net worth. That gap is the exposure — not the ticker itself.
The cost of concentration
What does concentration actually cost when it breaks?
It costs asymmetrically, and the math is unforgiving. A 50% drawdown requires a 100% gain just to return to flat — geometric drag, not arithmetic. In a real correction your diversifiers stop diversifying: correlations converge toward 1 exactly when you need them apart, so the “spread” you assumed evaporates in the same session. The NASDAQ shed roughly 80% from 2000–2002; Black Monday compressed a 22.6% loss into a single afternoon. We hold these failures in a structured crisis canon precisely so the mechanism is legible, not anecdotal. Now add a lock-up or restricted window and you hold a position you cannot exit while it reprices. Paper wealth is destroyed before liquidity returns.
Engineered resilience
What does engineered resilience look like?
It looks like a known failure surface. You stop guessing where the position breaks and start reading it — the drawdown that triggers a hedge, the regime that snaps your correlations, the cost of insurance carried before you need it. Velocity preserved, downside bounded, the math on the table.
The architecture
How it works
Three instruments, each pointed at one failure mode. You run them yourself — no call, no sequence.
The System Diagnostic
Maps your current tail exposure the way a profiler maps hot paths: where the position breaks, and at what magnitude.
The Fail-Safe Circuit
Programmatic tail-risk put-option protection that executes when correlation structure breaks down — your liquidity, insulated.
The Sandbox Engine
Compiles 50,000 extreme macro paths against your allocation. Don’t predict the crash — simulate it, edit the inputs, watch the distribution move.
Stated spec
The diagnostic carries a published 4.2% false-comfort rate — the share of runs that understate realized tail loss. We print it because a model that hides its error rate is the failure mode. The engine uses diffusion sampling over fat-tailed distributions rather than Gaussian VaR, which systematically underweights the tails (Spitznagel, Safe Haven, 2021); tails are estimated nonparametrically, not assumed. The full quantitative method and its limits are documented, and the structural case for tail-risk hedging sits one layer beneath this page.
Every financial crisis is a painting of human nature; we pair ten crises with ten artists to expose the structural failure beneath each one.
Audit it yourself
“Is this just an API wrapper?”
No — and you can break it to check. The sandbox is editable: change the inputs, re-run, and read the exposed equations behind each path. The stated limitation stands in view (4.2% false-comfort under fast-correlated shocks), and the simulation paths export. A wrapper hides its math; this one hands it to you to audit.
3 fields · 48-hour document · no call, no sequence.
Frequently asked questions
What is tail risk?
Tail risk is the exposure to rare, high-magnitude losses that standard models price as near-impossible but markets deliver anyway. For concentrated equity it is the dominant risk: the NASDAQ lost roughly 80% from 2000–2002, and Black Monday compressed a 22.6% drop into one afternoon. Gaussian VaR systematically understates it because real return distributions have fat tails.
Do I need a sales call to use the diagnostic?
No. You submit three fields and receive a 48-hour document; there is no call, no sequence, and no advisor in the loop. The diagnostic runs on your inputs, not on a conversation.
What does the diagnostic produce?
A known failure surface for your position: the drawdown that triggers a hedge, the regime that snaps your correlations, and the carry cost of insurance before you need it. It compiles 50,000 extreme macro paths against your allocation and ships the distribution, the exposed equations, and exportable paths — not a verdict.
Who is this for?
Operators and engineers holding 60–90% of net worth in a single ticker — typically founders, early employees, or technical leaders with concentrated equity. If your wealth is already diversified across uncorrelated assets, this is not built for you.
What does it cost, and when does it fail?
The diagnostic is run-it-yourself with no sales sequence attached. Its stated limit is a published 4.2% false-comfort rate — the share of runs that understate realized tail loss, concentrated under fast-correlated shocks. We print the failure rate because a model that hides its error is itself the failure mode.