- What if interest rates collapsed?
- What if energy prices doubled?
- What if the stock market fell thirty percent?
What if all three happened at once?
Aksel Fristrup, Founder of Henselian
Builds systems from the mathematics up
The problem
When asking another question costs too much, you stop asking.
- 01Insurers, banks and energy traders answer what-ifs with Monte Carlo simulation: millions of possible futures, each rolled out.
- 02A full run takes hours or days. So the scenarios are agreed in advance, and seldom changed.
- 03The combined shock, the new product, the question asked in the meeting: they wait for the next run, or never get one.
The solution
A simulation engine fast enough to ask again.
- 01The engineMonte Carlo on the GPU. Every member, loan or delivery day simulated on every path, with no model points.
- 02The languageHensel, the engine's own language. Models read like the mathematics and compile to the GPU.
- 03The appBuild a model in plain words, prove a migrated model agrees with the old one, and put each decision on record.
Measured, not claimed
Trillions of simulated paths. In seconds.
41s
1.5 trillion member-paths: a 2.5-million-member UK buy-in priced on 600,000 paths, end to end.
28,000×
faster than the Society of Actuaries' published reference on its nested variable-annuity benchmark.
Both measured on one machine with one NVIDIA RTX 4090. The buy-in is a synthetic scheme on an illustrative basis. The SOA benchmark in µs per policy-scenario: 789.7 for its vectorised NumPy reference, 0.0283 for Henselian.
From what if to what then
It is not just about asking what if. It is about answering what then.
What then should we
protect?
The seven points of coverage still standing above the early warning.
change?
The inflation exposure. Hedged, coverage climbs back above 180%.
do today?
Decide with the cost in front of you, and put the decision on record.
Illustrative example, not a forecast. The wall above the sea is the surplus.
Where it applies
One engine. Three industries.
Insurance
What if inflation surges through every claim?
Banking
What if markets move together?
Energy
What if prices swing unexpectedly?
Where we are today
The product is built. Now we are arranging the first pilots.
- BuiltEngine, language and appMeasured at full size on one GPU, for insurance, banking and energy.
- NowArranging the first pilotsWith teams that have a simulation problem they wish they could solve faster.
- NextPilots on real problemsRun first on a synthetic book shaped like theirs, then on their own data.
What we're looking for
Design partners to pilot. And introductions.
An insurer, bank or energy firm with a simulation problem it wishes it could solve faster. A pilot goes:
- 01We read it
- 02We model it
- 03We run it with you
Introductions to the people who own these questions: actuarial, risk, treasury and trading leads.
Bring us a problem: aksel@henselian.com
The greatest danger isn't always the scenario you calculated and got wrong.
Sometimes, it's the scenario you never had time to calculate.
Aksel Fristrup, Founder of Henselian
aksel@henselian.com · henselian.com