Cassandra Risking

What it is for

A prospect’s chance of success is a number that gets quoted in meetings for years and explained approximately once. By the time anyone asks why the seal risk was 0.6, the spreadsheet has been copied twice, the person who set it has moved on, and the reasoning exists only as a recollection. The number survives; the evidence behind it does not.

Cassandra structures geological risking so the evidence stays attached. Each assessment records the interpretation, the data it rests on, and the reviewer context — and a published scenario freezes all of it together, so the number can still be defended months later.

Geological chance of success, and only that

The scope is deliberately narrow. Cassandra calculates geological Pg and refuses to mix it with economics — no EMV, no recoverable volumes, no development chance. Those are real questions, but blending them into one figure is how a geological risk becomes an unfalsifiable commercial one. Keeping them apart is what lets a geologist defend the part they are actually responsible for.

How an assessment is structured

The classic components — Play, Source, Charge, Reservoir, Seal and Trap — each get their own assessment tab, with shared play risk kept explicitly separate from prospect and segment risk. That separation runs through the whole model:

• Play — the shared geological chance for the selected play: working petroleum system, regional seal preservation, migration fairway.
• Segment — the local interpretation currently selected, such as one mapped closure.
• Prospect — a roll-up of stored segment results within the active play, handling single- and multi-segment prospects.
• Cross-play — a review layer across plays for the same prospect, covering independent plays, overlapping plays and alternative concepts.

The evidence base

Every geological hypothesis can carry attached data and evidence alongside the judgment itself, with reviewer comments in the same place. Both judgment-led and evidence-native workflows are supported, because real assessments are a mix: some components rest on a measurement, others on an experienced opinion, and pretending otherwise makes the record less honest rather than more rigorous. Several datasets of the same type can coexist against one hypothesis, each labelled, so two seismic interpretations do not silently overwrite one another.

DHI, kept separate

Direct hydrocarbon indicators are handled as an explicit likelihood-ratio update on top of the geological assessment, rather than folded invisibly into a component. You can see what the base geological case was and what the DHI did to it — which is the only way a reviewer can disagree with one without discarding the other.

Publishing and reopening

Before a scenario is frozen, the app checks circulation and publish readiness. A published scenario becomes read-only and keeps its reproducibility metadata and review context, so the snapshot is genuinely a snapshot. Reopening it creates a new draft rather than editing the published record — the earlier decision stays intact and quotable, and the revision is visibly a revision.

Exports produce a risk narrative with that metadata attached, rather than a bare number.

What it does not do