One question, start to finish.
Ask what if; Henselian runs the model on every record and every path; you read what then.
- What if
What if people live longer than the basis says?
- Run
A buy-in price, member by member
Gompertz mortality improving each year, with a cohort shock on every longevity path; pensions discounted at gilts plus the matching adjustment.
- What then
The price, the BEL, and the BEL at 1-in-200 longevity: what to charge, and what to hold against it.
Start from a model that already asks.
Each is a working model in the app's Model builder, written in Hensel and run on the GPU. They start from a synthetic book; your data replaces it, record by record.
21 models in the Model builder
Bulk annuities4
A buy-in price, member by member
Price, BEL and 1-in-200 longevity BEL for a scheme, every member valued on every path.
Year-end BEL and SCR, every member simulated
BEL, the one-year SCR and the coverage ratio, every member revalued in every stress.
Capital by full nested simulation
A year of real-world risk, the whole book revalued on its own risk-neutral paths at its end: the 1-in-200 capital with no proxy.
A migration, reconciled member by member
Legacy and Hensel valuations side by side for every member, and where they part.
Life and pension3
The whole book, every member projected
BEL and the guarantee's cost on real-world paths, every member valued on every path.
Internal model SCR on the full book
The 99.5% one-year loss with rates, equities and longevity together, before and after the bonus potential.
Hedging the guarantees against the full book
The 1-in-100 guarantee loss with and without receiver swaptions, and what they cost.
P&C7
Motor renewal: expected cost, policy by policy
The baseline: each policy's expected claims given the year's frequency and inflation, no claims drawn. Scenario risk only.
Motor renewal: claims simulated, policy by policy
Every policy's claims drawn in every year, beside the expected-cost baseline on the same years: how much of the 1-in-100 is claim randomness.
Motor renewal: a premium for every policy
Each policy's indicated premium, its capped rate change and renewal premium, and who the caps leave under- or over-priced.
Windstorm losses, location by location
Each year's largest windstorm over every insured location: AAL, 1-in-200 and the 1-in-250 PML.
Reserves from every open claim
Best estimate, 75th percentile and the 99.5% tail excess of the ultimate cost, every open claim developed on every path by shared shocks.
Reserve risk, claim by claim: ultimate and one-year
Every open claim with its own development and settlement time, on the ultimate view and as a one-year claims development result.
A marine account, risk by risk
Attritional and port-accumulation losses on every bound risk, gross and net of a quota share and a per-risk excess of loss.
US statutory6
VM-20: indexed universal life, nested
An IUL block's net premium, deterministic and stochastic reserves, its index hedges priced by inner simulation at every future year.
VM-21: variable annuities under a hedging strategy, nested
A GMWB block hedged by its delta, the delta read every year on every scenario by inner simulation.
VM-21 hedging: Greeks and a lookahead hedge, nested three deep
A GMWB block hedged on its Greeks, read by inner simulation at every node, each year's hedge chosen on its own lookahead paths: CTE 70 hedged, delta-only and unhedged.
VM-22: fixed indexed annuities, nested
An FIA block's stochastic reserve, its hedge options priced by inner simulation at every future year, and the exclusion tests on the NAIC's 16 prescribed scenarios.
Policyholder behaviour: experience, dynamic lapse and the reserve
Actual over expected lapses by how far the guarantee is in the money, and what the assumed dynamic lapse does to the VA reserve.
Asset adequacy and C-3 Phase I: fixed deferred annuities
One cash-flow model of a deferred annuity block and its bonds: asset adequacy on the New York 7 and stochastic scenarios, and the C-3 Phase I interest rate capital.
Model validation1
Validating the scenarios and the book, record by record
Three independent checks: the martingale test, two halves of the paths against each other, and every member against a recomputation.
Fast enough to ask again.
A 2.5-million-member UK buy-in, every member on every one of 600,000 paths: 1.5 trillion member-paths in 41 seconds.
Replayed with both clocks sped up by the same 164×: the GPU bar fills in a blink; the CPU bar takes 34 seconds of your time.
One invitation
Come and challenge us.
If you work in insurance, banking, energy, or anywhere uncertainty drives important decisions, bring us the simulation problem your organisation wishes it could solve faster. Let's find out what's possible.