Strategy & Technology Consulting
Decisions need evidence.
JCLABS advises clients in finance, defence, and entertainment. We do not sell opinions. We build the evidence a decision rests on — and we write down what remains unproven.
Three sectors, one problem
The industries differ; the failure mode does not. Data sits in fragments, the reasoning behind a decision is never written down, and commitments get made on assumptions nobody tested.
Finance
Turning market data into evidence. We state a hypothesis, attack it, and keep only what survives.
- Strategy validation design
- Execution cost and fill analysis
- Risk limits and automatic stops
- Trading system audit
Defence
Making requirements testable, and leaving a written basis for every procurement decision.
- Technical feasibility review
- Acquisition rationale documentation
- Data architecture and operating procedure
- Test and evaluation criteria
Entertainment
Putting numbers on production and distribution — replacing instinct with measurement.
- Production pipeline automation
- Unit cost and recovery analysis
- Multi-channel distribution design
- Performance measurement systems
Built to be refuted
One side builds the case; another is paid to break it. Only what survives reaches the client. When a conclusion depends on a contested estimate, we set that estimate to zero and run it again.
Hypothesis and measurement method are fixed in writing first. We do not move the criteria after seeing the data.
An independent reviewer attacks the sample definition, clustering, multiple testing, and execution assumptions. Agreement without grounds is not accepted.
What survives ships as a reproducible procedure and code. What fails ships as a written failure record. Discarded hypotheses are kept, not deleted.
“If the upper bound still exceeds the cost, the answer is not ‘no’. It is ‘not yet known’.”
What every deliverable states
A result missing any of these is not treated as a result — internally or on client work.
- ControlA good number without something to compare it against is not evidence. We judge the difference from a control, never the level itself.
- ClusteringObservations from one event are not independent samples. Counting them as such inflates the evidence.
- PowerWe state up front how much data a verdict requires. When there isn't enough, we say there isn't enough.
- Multiple testingThe best result from a wide search is not taken at face value. It is re-tested once, on data held back for that purpose.
- FeasibilityWhat is possible in theory and what executes in practice are separated. We measure the constraint and size the design to it.
- Self-checkIf the expected value on randomised input is not zero, the code is wrong. That is checked before any result is read.
The method is applied to our own capital first
The verification discipline we bring to clients runs against our own systems before it runs against theirs. We operate a research programme on short-horizon price structure in digital asset markets — collection, hypothesis, refutation, and execution feasibility — end to end.
No live capital is deployed. The system records what it would have done without placing orders, and no funds are committed until an edge is shown to exceed its costs. The same threshold applies to client engagements.
Get in touch
Consulting enquiries, technical partnerships, and hiring.
- jchoi0125@gmail.com
- Phone
- —
- Office
- Singapore