PERISCOPE
External loss intelligence: other banks' losses turned into structured, scored, translated inputs.
The problem
Other banks' losses are the cheapest risk intelligence available, and most institutions waste them. An industry event makes the news, a summary circulates, someone says "could that happen here," and the conversation dies in the meeting where it started. Nothing is extracted, nothing is scored, nothing reaches the scenario program or the control conversation. External loss data ends up as anecdote with a subscription fee.
What it does
PERISCOPE turns external events into structured inputs. Each event is decomposed into a consistent shape: what failed, which controls were absent or defeated, which business lines and processes it touched, how the loss developed, and what the amounts were. Relevance scoring compares the event's profile against the bank's own footprint, so a payments-fraud event at a payments-heavy peer outranks a headline twice its size in a business the bank does not run. A could-it-happen-here pass translates the event into the bank's own terms: the equivalent process, the controls that would have to hold, and the honest answer about whether they would. Severity translation rescales external amounts to the bank's size and mix instead of quoting someone else's loss as if it were yours.
Every event in the demonstration library is fictional, written to be instructive, and tagged as fictional on screen.
Why it matters
Basel's principles have always listed external data as a core element of the identification toolkit, and every scenario regime expects severe-but-plausible narratives grounded in something. Structured external intelligence feeds both: scenario programs get starting points with evidence behind them, and control owners get the uncomfortable, specific question instead of the headline. The distance between "we read about it" and "we tested ourselves against it" is the entire value of the exercise.
Honest framing
PERISCOPE is a working prototype and a blueprint, not a product. It runs in the browser on a synthetic demo tenant with a fictional event library and deterministic demo scoring. Open it, pick the event that most resembles your nightmare, and read the could-it-happen-here answer. That is what it is for.